Mice were trained on a variety of odor detection and discrimination tasks in 100- or 200-trial sessions using a go, no-go discrete trials operant conditioning procedure. Odors, presented for 1 s on each trial, were generated by an air dilution olfactometer (for threshold tests) and an easily constructed eight-channel liquid dilution unit (for two- and multiple-odor discrimination tasks). Mice rapidly acquired the operant task and demonstrated excellent stimulus control by odor vapors. Their absolute detection threshold for ethyl acetate was similar to that obtained with rats using similar methods. They readily acquired four separate two-odor discrimination tasks and continued to perform well when all eight odors were presented in random order in the same session and when reinforcement probability for correct responding was decreased from 1 to 0.5. Memory for these eight odors, assessed under extinction after a 32 day rest period, was essentially perfect. Time spent sampling the odor on S+ and S- trials was highly correlated with response accuracy. When accuracy was at chance levels (e.g. initial trials on a novel task), stimulus sampling time on both S+ and S- trials was approximately 0.5-0.7 s. As response accuracy increased, sampling time on S+ trials tended to increase and remain higher than sampling time on S- trials.