In this paper, we revisit the problem of self-disconnection among prepayment energy customers. Using metering data from 2.3 million electricity pre-payment customers, we study how often households with an electricity pre-payment meter tend to self-disconnect over the course of a year - and why they do so. What we find is that, in any given year, the majority of households (ca. 78%) do not self-disconnect; ca. 12% self-disconnect once; ca. 3% self disconnect more often than four times. We also find that most self disconnections (ca. 62%) last for less than one day; between 72% and 82% last for less than two days; 12%-18% last for more than 3 days. As for the main driver of self-disconnection, we identify financial constraints. This suggests that it is likely to be difficult/expensive to reduce the total number of self-disconnections. In the last part of the paper, we argue, however, that it may (still) be possible to reduce the negative impact of self-disconnection in a relatively inexpensive way - at least to some extent - by helping households to better smooth their self disconnections over the course of a year.