Abstract Response latencies obtained from a computer administered version of the Marlowe-Crowne social desirability scale (MC) were examined. Not unexpectedly, reading speed accounted for a major portion of the variance in response latency. However, after controlling for reading speed, the proportion of individuals in the sample endorsing an item on the MC and the item-total correlation contributed unique variance. Latencies for true responses to items on the MC were found to be quicker than for false responses for respondents in two separate samples. In the larger study ( N=169), there was a significant interaction between true/false response and scoring direction with true responses to items considered socially desirable if answered false having the shortest average response latency and false responses to items scored as socially desirable if answered true having the longest average response latency. The second sample of participants ( N=40) had shorter response latencies for true responses and for items scored as socially desirable if answered true; there was not a significant interaction.