Depending on exactly what device you're checking, it can be something as simple as a common household oven thermometer (for checking tempering ovens), to building one of these....
http://www.caffreyknives.net/forge_pyrometer.html
If you mean checking for accuracy in a heat treat oven, just be aware that it likely won't be what you expect. Different brands of ovens are made to different tolerances in terms of percentage +/- of the set point. In general, the more expensive heat treat ovens range in the 2-4% accuracy from set point, and that gap usually gets larger the cheaper the oven. To explain, I'll use the temp 2000F. If a given heat treat oven has a 3% variance tolerance, the oven is considered to be within tolerances if the temp varies 60 degrees either way of a 2000F set point. If the oven's tolerance range is 7%, then it's acceptable for the temp to range 140 degrees either way of a 2000F set point. Those tolerance percentages are generally why you can find such varying prices on the same sized heat treat ovens we see in the knifemaking world.
The only way I found out about this was because I owned a heat treat oven that varied much more then I thought it should, and after going back and forth with the company, assuming that I had a "bad" oven (and expecting them to replace it), one of the tech support individuals explained it to me. As it was, the oven in question was one of the cheaper versions/brands that I had purchased based on it being the least expensive. Turns out it had a 9% tolerance range. I sold that one off and bit the bullet
About the only way you will ever get a heat treat oven that is "dead on", is to purchase a "Laboratory Grade" oven. These particular ovens are built to the highest possible standards, and often start at $10K+.