The correlation between Meteosat water vapour (WV) data and positive anomalies of potential vorticity (PV) is investigated using a case study involving cyclogenesis in the Alpine region and a north-west African depression. Correlation analyses are performed at different isobaric surfaces in the middle and upper troposphere for two latitudinal belts (25–35° N – ‘southern’, and 35–52.5°N – ‘northern’). A significant correlation is obtained at 500 hPa (correlation coefficient range: 0.58–0.64). Linear regression models are fitted to the samples from the northern and southern belts. On the basis of these results, the use of WV data as a tool for detection of errors in model-derived PV fields is discussed. It is concluded that, though some of the discrepancies between PV fields and WV images could be due to errors in the model analysis or prognosis, some mismatches are possibly due to the variable sensitivity of the WV channel because of the changes in pressure over the commonly used PV fields. It is suggested that the comparison of WV data with the field of positive PV anomalies at 500 hPa might be another approach to help solve the problem of verifying and validating numerical output.