Periodically the ARRL conducts an on the air Frequency Measurement Test. They have a couple of different formats they use, but essentially they challenge contestants to measure several frequencies accurately. In recent times one of our local group members has worked with the ARRL to transmit some of the test frequencies from the SF Bay Area since propagation makes hearing Newington, CT difficult at times.
The new web entry format allows the contestants to post their entries on a website, which is very convenient and speeds up the whole process. The results are made public about 14 days after the test.
Other frequency measurement tests are conducted by different folks that allow more frequent practice and equipment testing.
The recent ARRL test was November 13, 2008 0245 GMT which put it on Wednesday November 12 in the early evening here for us on the West Coast.
The equipment required to make a credible entry in this test is NOT substantial. If you look through the contestants in the last several tests that placed within one or two hz the equipment is often quite simple. Common transceivers that are not noted for stability are well represented such as FT817, IC706 and the K2. You don't need a Cesium, Rubidium, GPS timebase, precision synthesizer or anything else fancy. A basic radio and a computer with some free software seem to be quite adequate.
The basic idea is to use a program in the computer that analyzes the audio input from the sound card. A single cable from the computer to the radio will suffice to get sound into the computer. No fancy interfaces are required since you are not going to transmit - RF isolation and PTT are not needed.
3) Move to the measurement frequency. Use the radio to tune the incoming frequency to nearly the same frequency used during calibration. Record the radio settings and read the frequency of the incoming audio as accurately as you can. Take several good readings for averaging later. Note the quality of the readings as this will be helpful later. In some cases there will be more than one frequency to measure, do them with the radio at the same settings if possible (frequency and mode). Move on to the next band or station and record those measurements. Make plenty of notes as you go to clarify things later during the analysis. (It may be helpful to set up a paper form or spreadsheet to enter the data in an organized fashion).
NOTE - for the recent style of ARRL FMT the only values entered are the frequency differences, so absolute calibration is not required. Just read the carrier and two tone frequencies from the FFT program and subtract them to find the differences. Much simpler.
5 MHZ: 4.999 MHZ + 1.011 KHZ; 2.2e-6 error
NOTE - modern radios use synthesizers to create various frequencies, and have varying precision. This causes the radios to tune in uneven steps. These steps will be anywhere from a few millihertz to a few hz. Test your radio with WWV to see how big the actual tuning steps are and whether they vary.
NOTE - if you want to go to the next level do some research on the internet. A GPS Disciplined Oscillator is not expensive, and an accurate millihertz Synthesizer )available surplus at reasonable prices) can give you a precise frequency reference within the receiver passband so the receiver's tuning steps are not a problem. Then the task becomes measuring the difference between the accurate reference and the unknown signals.
NOTE - measuring signal frequencies using oscilloscope matching of lissajous patterns or matching waveforms works well for constant long term signals, but for these short tones the FFT technique is required. The short tones just don't last long enough. If you record the session the analysis can be repeated and reviewed for additional learning and precision.

Other Ideas
Using CHU which is closer to the ham band than WWV is another idea.

Sample Calculations:
The calculations are fairly simple, but it is easy to get confused and subtract instead of adding. Remember that the sideband used changes the addition/subtraction that needs to be done. For example set your receiver to 9.999 mhz USB (upper sideband) and find the carrier frequency with your audio program. It should be near 1 khz. The sum of the two values should be 10.0 MHZ, the carrier of WWV. Now change to LSB (lower sideband) and change to 10.001 MHZ. The WWV carrier should again be at about 1 khz, but this time the audio frequency must be subtracted from the receiver frequency to get the proper 10.0 MHZ value. Now change to CW mode and set the rig frequency to peak the carrier in the CW filter. Now read the radio's readout and the audio frequency from the program. Different radios do different things in CW mode, so you will have to figure out whether it is USB or LSB, and whether it has an offset for the cw offset. Many or most rigs read the transmit frequency in CW mode, and the receiver is offset by the CW filter center frequency. For this reason frequency determination is a bit more complicated in CW mode. For the contest it may be easier to stick with USB since this keeps the math simple and there is no offset to worry about.
The other aspect of this is calibration. When you measure the carrier frequency of WWV, does it come out perfectly on the known frequency? It is likely a little off. Now you could disassemble the radio and attempt to calibrate it, but that is not required. If it is easy to calibrate the rig, and you want to do that, go ahead. Otherwise, measure WWV at several frequencies and see if the error is proportional to the carrier frequency. If so, it is easy to mathematically calibrate it out. By proportional, is the error on 10 mhz twice the error on 5 mhz? If so, calculate a percentage error, and apply this to the measurement in reverse to get a more accurate reading.
Let's do a simple example. Let the rig warm up and then check our receiver in USB mode with WWV at 5, 10 and 15 MHZ by setting it 1 KHZ low in frequency and reading the audio frequency with the computer soundcard and software. We get the following results:
5 MHZ: 4.999 MHZ + 1.011 KHZ; 2.2e-6
10 MHZ: 9.999 MHZ + 1.023 KHZ; 2.3e-6
15 MHZ: 14.999 MHZ + 1.032 KHZ; 2.133e-6
Calculate the calibration error by subtracting the sum from the correct frequency and dividing by the correct frequency. In this case the error is about 2.2 parts per million, and the reading is high. So when we correct later we need to subtract the error from the apparent frequency to get the actual frequency.
Now let's do a measurement of a signal on the air. Tune the signal for about 1.0 khz and read both the receiver readout and the audio frequency:
unknown signal apparent frequency = 7.04512 MHZ + 1.01335 KHZ = 7.04613335 MHZ
Now we know the rig is reading about 2.25e-6 high (averaging the 2.2 and 2.3 values that are above and below the frequency of interest). So we need to subtract 15.854 HZ from that resulting in:
error = 7.04613335 MHZ * 2.25e-6 = 15.854 HZ
actual unknown signal frequency = 7.04613335 MHZ - 15.854 HZ = 7.046117496 MHZ
Using a spreadsheet is probably the easiest way to do this. The calibration of the sound card is not very important if we always measure close to the same frequency with it.
One other interesting idea that was used by some was to capture the audio and settings using a video camera!

0) Load an audio FFT program and cable the audio from the radio to the sound card. It is convenient to also have a speaker or phones available to listen to the audio. Perform some test runs on WWV on several bands (use one WWV frequency as the reference and measure WWV on another frequency such as 5, 10 and 15 mhz). Familiarize yourself with the software and the data recording process.
1) Warm the radio up thoroughly (say a few hours).
2) Tune to the WWV frequency above (or below, or do both) the frequency to be measured, using the same USB or LSB or CW setting that you plan to use to measure the unknown signal, tuning such that WWV's carrier frequency is about where you will put the unknown signal (say the center of the CW filter). Record the radio settings (Frequency, mode, sideband, etc). Measure the audio frequency of the WWV carrier with the audio program you have selected.
3) Move to the measurement frequency. Use the radio to tune the incoming frequency to nearly the same frequency used during calibration. Record the radio settings and read the frequency of the incoming audio as accurately as you can. Take several good readings for averaging later. Note the quality of the readings as this will be helpful later. In some cases there will be more than one frequency to measure, do them with the radio at the same setting if possible. Move on to the next band or station and record those measurements. Make plenty of notes as you go to clarify things later during the analysis. (It may be helpful to set up a paper form or spreadsheet to enter the data in an organized fashion).
There are many audio programs that are suitable for accurately reading the frequency of the tones in the passband. Digital mode programs are often used as well as audio analysis software. Look through the contestants in previous FMTs to see what they are using.
Review articles on FMT techniques:

DL4YHF Audio Spectrum Lab∞
The ARRL FMT Main Web Page∞
Review recent FMT results∞
Enter Your FMT results∞

ARRL FMT = Frequency Measurement Test

The ARRL FMT Web Main Page∞
Take a look at recent FMT results
Enter Your FMT results

0) Load an audio FFT program and cable the audio from the radio to the sound card. It is convenient to also have a speaker or phones available to listen to the audio.

The basic plan:
Further Reading:
Enter FMT results at this website:
Review some articles on FMT techniques:
73,