In this study, we apply flexible Fourier stationary unit root test proposed by Enders and Lee (2012) to assess the non‐stationary properties of the per capita real gross domestic product (GDP) for 32 African countries. We find that Fourier stationary unit root test has higher power than linear method if the true data‐generating process of per capita real GDP is in fact a stationary nonlinear process of an unknown form with structural change using the low frequency components. We investigate the stationarity of per capita real GDP from the nonlinear point of view and provide robust evidence that clearly indicates that real output is well characterised by a nonlinear, mean‐reverting process, namely Benin, Botswana, Burundi, Cameroon, Senegal, Sierra Leone and South Africa. Our evidence points that these seven countries are nonlinear stationary, implying that per capita real GDP follows a steady rate of growth, and policy innovations then have temporary effects. These results have important policy implications for African countries.