Drug effects on human memory are usually assessed by overt recall or recognition tests. Covert tests which do not explicitly assess memory but which indirectly elicit previously presented information may be more sensitive to low levels of learning than overt tests. Three covert tests and corresponding overt recall tests were given to 16 men and 16 women breathing 30% nitrous oxide in oxygen or 100% oxygen to see if the covert tests resisted the memory-impairing effects of nitrous oxide. Nitrous oxide impaired performance on the overt tests. Performance in two covert tests, Constrained Associations and Word Completion, showed resistance to memory impairment. In the least resistant covert test, Free Associations, nitrous oxide altered associative processes. Performance in an additional test involving recognition and preferences for nonsense words repeated with varying frequencies also showed some resistance to memory impairment. The results support the distinction between declarative and procedural memory. Constrained Associations, Word Completion, and Nonsense Words tests may be useful for assessing low levels of learning during drug states.