South Africa’s Living Conditions Surveys (LCSs) are crucial instruments in monitoring the well-being of South Africa’s population. Using expenditure as the yardstick for assessing well-being, the LCSs show a marked drop in poverty between 2008/09 and 2014/09. This would be a welcome trend. Unfortunately, the data collection method changed between these surveys so that there is at least some doubt as to whether the trend is real or due to measurement changes. Given that the expenditure modules are quite onerous for respondents this is an important question. In this paper we assess the reliability of the measured improvement of welfare by analysing also the information in the asset modules of these surveys. In comparison to the expenditure diaries the asset modules are much easier to complete and should therefore be much less prone to measurement error. We combine the information in the asset variables by constructing various asset indices and show that if these are used across the surveys they confirm that indeed average well-being has improved between 2008/09 and 2014/15.