Two experiments were performed to investigate the effects of transmission delay and access delay, respectively, on the efficiency with which speakers verbally encoded information for transmission in a two-person communication task. Both experiments employed echo-free four-wire voice circuits in an attempt to isolate each delay effect and to avoid the delayed echo effect found in commercial circuits. In the first experiment, three values of pure roundtrip transmission delay were used: no delay, 0.6 sec, and 1.8 sec. Using 14 pairs of male subjects in each condition, it was found that, whereas 1.8 sec of transmission delay deleteriously affected the efficiency of communication, subjects performed as efficiently using the 0.6-sec delay circuit as with no delay. In the second experiment, three values of access delay were used: no delay, 0.25 sec, and 1.8 sec. Ten pairs of male subjects and 10 pairs of female subjects were run in each condition. The effect of access delay was found to be different for the two sexes. Access delay of 1.8 sec had a greater effect on males than on females, whereas at 0.25-sec delay female performance was impaired slightly and male performance not at all. With no delay, male and female performance did not differ. These results are supported by data based on subjects' responses to a postexperiment questionnaire. The findings are discussed in terms of their relevance to problems encountered in two-way voice communication over long transmission paths.