When respondents are asked to report on past behaviour, their responses may be affected by an unknown level of measurement error. This casts doubt on the findings from retrospective surveys. There is evidence that measurement error is a function of the interval between an experience and the time when the experience is reported. In this study, the volume of word of mouth (WOM) is measured as a function of this interval. Both positive and negative WOM (PWOM, NWOM) show little change with interval, which indicates that recall measures of the volume of WOM are quite reliable and may be used with confidence. Possible distorting influences on retrospective measures are discussed.