Ron's absolute right, of course. However, I was speaking of the typical DMM out there, especially the cheaper ones. This is a good opportunity to present a bit more information.That isn't quite true. If you look closely at the image of the Fluke DMM you will see a REL Δ button which is used to zero out the meter. The .1 ohm shown in the image was the lead resistance in the measurement plane. However, pressing the relative button subtracts the lead resistance from the actual reading. So in the case of higher end DMMs it is possible to read low resistances.
First, lemme give a good example of what Ron's talking about. My HP bench meter with some leads measures my GR resistance box set to 0.5 ohms resistance as 0.638 ohms. When I short the leads, I get 0.123 ohms. Making the correction, I get the resistance of 0.515 ohms. This is quite close to the 0.5165 ohms I measured using my 1 A current source.
However, low resistance readings with hand-held digital multimeters can have poor accuracy, even when you correct for the lead resistance. For example, my Fluke 83's (a 20 year old meter) accuracy for resistance measurement on the lowest scale is ±(0.4% of reading plus 1 digit). Thus, if I read a resistance of 0.5 ohms (with the lead resistance subtracted out), the accuracy is (essentially) ±0.1 ohms. That's a 20% uncertainty in the reading, so it's something you have to be aware of (and it's due to the digital uncertainty in the last digit). If you were using an Agilent U1252B, which can read to 0.01 ohms, the accuracy spec is ±(0.05% + 10) (this spec also requires you to use the null technique to get rid of lead resistances). To repeat the same example 0.5 ohm measurement, you'd see that the 10 digits part of the spec would result in the same accuracy as the Fluke. Thus, the Agilent meter that looks much better in fact has the same (poor) 20% of reading accuracy for this particular measurement.
If you measure low resistances a lot like I do, it's worth your time to build a current source to make a Kelvin measurement. For example, the CurrentSource.pdf file here gives an example. Another common approach is to use a current meter to set a constant current DC power supply to a particular current, then use it as a constant current source. But building a dedicated current source will save you time. It will also greatly improve the accuracy of your resistance measurements because the accuracy will be based on the DC measurement accuracy of your meter (assuming you calibrate the current source properly). From the above arguments, you can see the accuracy for measuring resistances with a hand-held digital multimeter at an ohm and below will be 10% or substantially larger; measuring the DC voltage drop of the current source will instead let you use the 0.1% or better accuracy of the DC specs of your meter.
There's another issue with resistance measurement to be aware of. When you measure low voltages, you'll often be bedeviled with thermoelectric interference. The same types of voltages can also cause errors when making resistance measurements. They're insidious because, depending on their polarity, they can either add or subtract to the measured resistance. And, if you're not aware of them, they are pure errors in your measurement.
A similar problem can occur when you measure resistances in a circuit. If you've ever measured a negative resistance, then you've experienced what I'm talking about. It's caused by a voltage present in the circuit whose resistance you're trying to measure. Digital multimeters like my HP bench meter (and follow-on Agilent meters) have offset-compensated resistance measurements. Basically, the meter measures the circuit for a voltage, then applies the current source and measures the resistance; the first voltage measurement is used to correct the offset in the second measurement. Naturally, your reading rate halves. Other meters measure the resistance, then interchange the leads and measure again. The measured resistance is then the average of the two readings (this subtracts out the effect of the offset voltage).
This brings up a technique that perhaps not too many folks know. When you're measuring resistances in an unknown circuit, it's wise to flip your leads and measure again. A difference will tell you there's an offset voltage in the circuit. You can average your readings to get the true resistance. This can save you from making a mistake some day. For both this technique and measuring diode drops, I've long wished I had a set of leads that had a small button on one lead to make this lead interchange at the press of the button...