The current study (n = 180) assesses factors affecting the recall of past mobile phone behaviour. It manipulates two factors hypothesised as influencing recall of duration and number of calls made: time reference (24 hours vs. 3 days) and context prompt (temporal, person called, reason for call) and also considers their impact in relation to levels of mobile phone use. A measure of recall error was constructed by matching self-reported calls to actual calls and weighting for actual usage. The results indicate that most people tend to overestimate duration and underestimate number of phone calls, although these discrepancies are related to actual amount of mobile phone use. The manipulations of time reference and context prompt have some effect on the patterns of recall in relation to number, rather than duration, of calls. The implications of these results for the development of reliable and valid self-report measures of mobile phone use are discussed. Copyright © 2008 John Wiley & Sons, Ltd.