Objective: We wanted to determine the true rate of chance responding on the Word Memory Test (WMT) and Test of Memory Malingering (TOMM). The assumption is that forced choice pairs are 50/50 binomial probabilities. Method: 55 undergraduate participants were given the WMT immediate recognition trial and TOMM trial 1 [mean age 21.9 years (SD 5.9), 64% white, 29% African American, 53% female]. A sham condition was used to induce random responding on the WMT and TOMM where the participant was instructed to telepathically determine the answer chosen by a responder in another room for each response pair on the WMT and TOMM. The sham responder in the other room was either a person or a computer for different scenarios and was counterbalanced across participants. Results: The mean rate of correct responding for participants who were randomly responding was 60% (SD .09) on the WMT and 52% (SD .07) on the TOMM. One sample t-tests showed that both of these conditions were significantly different from 50% [WMT t(54) = 8.311, p < .001; TOMM t(54) = 2.166, p = .039]. Conclusion: Our results suggest that the forced choice trials of the WMT and TOMM are not 50/50 probabilities as is commonly assumed. We believe that the semantic relationships between the forced choice pairs result in greater than 50% correct response rates in individuals who are randomly picking a response. This semantic effect appears to be particularly strong for the WMT with random responding resulting in a 60% correct response rate on this test.