This paper investigates asymmetric cointegration and causality effects between financial development and economic growth for South African data spanning over the period of 1992 to 2013. To this end, we make use of the momentum threshold autoregressive (MTAR) approach which allows for threshold error correction (TEC) modelling and granger causality analysis between the variables. In carrying out our empirical analysis, we employ six measures of the financial development variables against gross domestic per capita, that is, three measures which proxy banking activity and another three proxies for stock market development. The empirical results generally indicate an abrupt asymmetric cointegration relationship between banking activity and economic growth, on one hand, and a smooth cointegration relationship between stock market activity and economic growth, on the other hand. Moreover, causality analysis generally reveals that while banking activity tends to granger causes economic growth, stock market activity is, however, caused by economic growth increase.