I don't doubt that K-Ar dating of rocks is a sound method, but in discussions with skeptics it would be good to know just how reliable exactly? For example, has K-Ar been cross-checked with other rock dating methods, such as radiocarbon, sediment varves, etc. If so, yielding what level of correlation?
The reliability is dependent on the quality, (for instance is the mineral reworked or porous and thus prone to contamination), and size of the sample, (small samples result in poorer measurement). There is also of course the date range issue, use a sample that is too old or too new and the amount of each element is so tiny it can't accurately be measured. Thats why multiple measurements are always taken, if they vary wildly you know something is wrong. A good general rule for modern studies is expect a date to vary by ±5%.
Potassium-argon can be compared with relative dating methods like fossils and bedding as well as absolute dating methods like Uranium-Thorium, Uranium Lead, Rubidium Strontium, and even it's own dates to check for consistency. This is always being improved upon and multiple dates from different methods are sought whenever possible so there are tens of thousands of comparisons.
There is a good article about it here. They also provide a good example study (under the title a good example) which compares several methods for dating the K-T boundary and yields an error of about ~1 million years between them . Beware of older studies (1970's or older) which can have much larger error bars.
For other studies comparing lots of dates your best bet is large scale timescale reconstructions like this one which will compare dozens if not hundreds of individual dates.