So what DO they really mean in the real world?
I've had my own idea for several years as to what they mean to me:
S-1 - Faint signal, barely perceptible: This could be a signal so weak that you only hear a carrier, no modulation at all, or something where you hear modulation and can distinguish some programming, but with considerable difficulty. For me, a RS report of 51 (perfectly readable) will almost NEVER happen, and a 41 (readable with practically no difficulty) would be very rare, usually only achievable if I have no other audio sources of noise/sound nearby, and am listening on sound-isolating headphones. Reports of 31 and 21 would be most common for me, with 11 (unreadable, faint signal barely perceptible) being reserved for situations where I could tell a signal is there, but can't understand any modulation. (While I'm also on the subject of readability, it is also possible for me to report a signal as, say, 2-9, or 1-9, although this would usually only happen extremely close to the transmitter site where my receiver is being severely overloaded.)
S-2 - very weak signal: These would nearly always get readability reports of 5, with an occasional 4. (That's assuming the transmitting station doesn't have any issues with its audio processing.) The signal may still seem to be somewhat buried under the noise, but could definitely be heard.
S-4 - fair signal: (S-3 would be basically between a S-2 and a S-4) A signal of this strength would usually light the tuning indicator on a radio that uses it, or trip the scan function on other radios so it will stop on that frequency.
S-5 - fairly good signal: would still have some audible static in it, but the signal would be fairly well above the noise. As an example, my reception here in El Cajon, CA, of KNX-1070 (50kW, 111 miles, part saltwater path) and KFI-640 (50kW, 99 miles, mostly land path) would be rated as S-5 by me on most portable radios (although I may consider rating them as S-6 on my Tecsun PL-380.)
S-6 - good signal: Any sign of static would be faint, and if music or similar programming is being played on the frequency the noise would be completely masked.
S-7 - moderately strong signal (although a Ham book I read when I was younger defined this as "strong signal"): There will be virtually NO trace of noise on the channel, even on a signal carrying talk programming - signal-to-noise ratio may exceed 60dB or so, but I haven't measured or experimented to find out for sure. Many local stations here (KOGO-600, XEWW-690, KFMB-760, KECR-910, KSDO-1130, KCBQ-1170 and KLSD-1360) fall into this or higher categories. (If you have a source of noise nearby - one that's capable of completely wiping out any trace of the earlier referenced reception of KNX and KFI, then you may hear noise on these S-7 signals.)
Signals at S-7 would not be defined by the apparent signal to noise ratio, but by the overload characteristics of the receiver. For example, on my Panasonic RQ-SW10, a radio with poor selectivity like the Sony SRF-M37W, a signal with an S-7 strength will sound just as loud at +/-10kHz as it does on the frequency, and will be lighting its tune indicator off-channel.
S-8 - strong signal (the Ham book I read many years ago reported this as "very strong signal"): As with S-7, there will be no trace of static. Splatter off-channel will be a little wider than a S-7, with the station typically heard about +/-20 or +/-30kHz just as loud as on the fundamental on poor-selectivity radios, and probably +/-50 to 70 or more at progressively reduced levels. On some other radios, for example my DSP-based Tecsun PL-380, the RSSI indicator would indicate a signal somewhat over 40dBuV about 20-30kHz, maybe a bit more, on either side of the station's frequency.
S-9 - extremely strong signal: Almost NO source of noise is capable of "jamming" this signal, unless your radio AND the noise source are both in the same Faraday cage. "Loud" splatter would occur AT LEAST +/-30 to 50kHz on either side of the assigned frequency on a poor-selectivity radio, with the signal being faintly heard on other frequencies over a majority of the band in many cases. The aforementioned Tecsun PL-380 would be displaying a RSSI of 49 to 50dBu off-channel over a fair portion of the band (possibly 100kHz or more on either side of the station's frequency, in many cases across most of the band.) For example, when their transmitter site was in Santee at the site where a Lowe's and Kohl's store now are, KCBQ-1170 was the ONLY station that I would have reported as a S-9, and it was about 5 miles or so away. As another example, my grandparents in San Gabriel, CA, are about 1/3 mile from the site of 23kW KAZN-1300 and 50kW KMRB-1430, and I would absolutely give those two stations S-9 reports at their house. The one other station that would also get a S-9 from there is 50kW KDIS-1110, whose transmitter is a little under 5 miles away from there. Generally, a report of S-9 will only happen if you are within about 5 miles of a 50kW transmitter over a path with fairly good ground conductivity (usually 10 or higher), with shorter distances required for less powerful stations, and longer distances allowed if the station is directional with a lobe aimed that direction, or over a saltwater path. I would guess that field strengths of S-9 signals would almost, if not always, exceed 500 millivolts/meter or so.
Now... that's basically how I've been personally interpreting them. In summary, anything below S7 would be interpreted based on the audible signal to noise ratio, with S7 and above being interpreted by the front-end overload characteristics of the radio (how wide it splatters on a poor-selectivity radio, or how wide it desenses a radio with better selectivity and to what level, as indicated by the signal strength meter - for example, on my PL-380 a reading of 50dBu would be considerably more severe desensitization than a reading of 30dBu.) However, I've sometimes heard of reports of S9+60dB in some cases. Using my interpretation, a signal like that would ALWAYS be COMPLETELY overloading your radio, and you'd probably have to be at their transmitter site to achieve it.
So... since it appears my interpretations may be wrong.... how are the signal strength reports really to be interpreted?
I've had my own idea for several years as to what they mean to me:
S-1 - Faint signal, barely perceptible: This could be a signal so weak that you only hear a carrier, no modulation at all, or something where you hear modulation and can distinguish some programming, but with considerable difficulty. For me, a RS report of 51 (perfectly readable) will almost NEVER happen, and a 41 (readable with practically no difficulty) would be very rare, usually only achievable if I have no other audio sources of noise/sound nearby, and am listening on sound-isolating headphones. Reports of 31 and 21 would be most common for me, with 11 (unreadable, faint signal barely perceptible) being reserved for situations where I could tell a signal is there, but can't understand any modulation. (While I'm also on the subject of readability, it is also possible for me to report a signal as, say, 2-9, or 1-9, although this would usually only happen extremely close to the transmitter site where my receiver is being severely overloaded.)
S-2 - very weak signal: These would nearly always get readability reports of 5, with an occasional 4. (That's assuming the transmitting station doesn't have any issues with its audio processing.) The signal may still seem to be somewhat buried under the noise, but could definitely be heard.
S-4 - fair signal: (S-3 would be basically between a S-2 and a S-4) A signal of this strength would usually light the tuning indicator on a radio that uses it, or trip the scan function on other radios so it will stop on that frequency.
S-5 - fairly good signal: would still have some audible static in it, but the signal would be fairly well above the noise. As an example, my reception here in El Cajon, CA, of KNX-1070 (50kW, 111 miles, part saltwater path) and KFI-640 (50kW, 99 miles, mostly land path) would be rated as S-5 by me on most portable radios (although I may consider rating them as S-6 on my Tecsun PL-380.)
S-6 - good signal: Any sign of static would be faint, and if music or similar programming is being played on the frequency the noise would be completely masked.
S-7 - moderately strong signal (although a Ham book I read when I was younger defined this as "strong signal"): There will be virtually NO trace of noise on the channel, even on a signal carrying talk programming - signal-to-noise ratio may exceed 60dB or so, but I haven't measured or experimented to find out for sure. Many local stations here (KOGO-600, XEWW-690, KFMB-760, KECR-910, KSDO-1130, KCBQ-1170 and KLSD-1360) fall into this or higher categories. (If you have a source of noise nearby - one that's capable of completely wiping out any trace of the earlier referenced reception of KNX and KFI, then you may hear noise on these S-7 signals.)
Signals at S-7 would not be defined by the apparent signal to noise ratio, but by the overload characteristics of the receiver. For example, on my Panasonic RQ-SW10, a radio with poor selectivity like the Sony SRF-M37W, a signal with an S-7 strength will sound just as loud at +/-10kHz as it does on the frequency, and will be lighting its tune indicator off-channel.
S-8 - strong signal (the Ham book I read many years ago reported this as "very strong signal"): As with S-7, there will be no trace of static. Splatter off-channel will be a little wider than a S-7, with the station typically heard about +/-20 or +/-30kHz just as loud as on the fundamental on poor-selectivity radios, and probably +/-50 to 70 or more at progressively reduced levels. On some other radios, for example my DSP-based Tecsun PL-380, the RSSI indicator would indicate a signal somewhat over 40dBuV about 20-30kHz, maybe a bit more, on either side of the station's frequency.
S-9 - extremely strong signal: Almost NO source of noise is capable of "jamming" this signal, unless your radio AND the noise source are both in the same Faraday cage. "Loud" splatter would occur AT LEAST +/-30 to 50kHz on either side of the assigned frequency on a poor-selectivity radio, with the signal being faintly heard on other frequencies over a majority of the band in many cases. The aforementioned Tecsun PL-380 would be displaying a RSSI of 49 to 50dBu off-channel over a fair portion of the band (possibly 100kHz or more on either side of the station's frequency, in many cases across most of the band.) For example, when their transmitter site was in Santee at the site where a Lowe's and Kohl's store now are, KCBQ-1170 was the ONLY station that I would have reported as a S-9, and it was about 5 miles or so away. As another example, my grandparents in San Gabriel, CA, are about 1/3 mile from the site of 23kW KAZN-1300 and 50kW KMRB-1430, and I would absolutely give those two stations S-9 reports at their house. The one other station that would also get a S-9 from there is 50kW KDIS-1110, whose transmitter is a little under 5 miles away from there. Generally, a report of S-9 will only happen if you are within about 5 miles of a 50kW transmitter over a path with fairly good ground conductivity (usually 10 or higher), with shorter distances required for less powerful stations, and longer distances allowed if the station is directional with a lobe aimed that direction, or over a saltwater path. I would guess that field strengths of S-9 signals would almost, if not always, exceed 500 millivolts/meter or so.
Now... that's basically how I've been personally interpreting them. In summary, anything below S7 would be interpreted based on the audible signal to noise ratio, with S7 and above being interpreted by the front-end overload characteristics of the radio (how wide it splatters on a poor-selectivity radio, or how wide it desenses a radio with better selectivity and to what level, as indicated by the signal strength meter - for example, on my PL-380 a reading of 50dBu would be considerably more severe desensitization than a reading of 30dBu.) However, I've sometimes heard of reports of S9+60dB in some cases. Using my interpretation, a signal like that would ALWAYS be COMPLETELY overloading your radio, and you'd probably have to be at their transmitter site to achieve it.
So... since it appears my interpretations may be wrong.... how are the signal strength reports really to be interpreted?
I was aware of the 50uV spec, but don't know much else.