Despite the fact that the original study had not been designed as a panel study (and despite the fact that no attempt to track the respondents had been made in the intervening four years), the response rate was (as we argue below) reasonably good. This paper describes the two waves of this Khayelitsha panel study, and provides an analysis of attrition. The first wave of the data (KMP 2000) is publicly available in the Data First Resource Centre of the CSSR.