River gaugings at Mohembo started on a more or less regular basis in 1974. Data of variable length are available on the following parameter: stage, discharge (gauged or estimated from rating curve), cross-sectional area, surface width, mean velocity, mean depth and maximum depth. There is, though, concern about the existence of 'inexplicable' outliers in the stage-discharge plot. This could be due to human error in the recordings of data. The aim of this study, therefore, is to investigate the reason for this instability and how a useful rating curve can be extracted from the available data.