This study examines the nonlinearity and chaotic behavior of the time series of returns of two exchange traded funds (ETFs) listed in Hong Kong Stock Exchanges, namely Hong Kong Tracker Fund (HKTF) and iShares FTSE A50 (ISFT), and the adequacy of autoregressive-generalized autoregressive conditional heteroskedasticity (AR-GARCH) models to capture nonlinearity. A set of nonlinearity tests consistently indicates the presence of nonlinearity in both return time series and the Brock–Dechert–Scheinkman (BDS) test of nonlinearity on AR-GARCH residuals, and the inability of AR-GARCH models to capture the nonlinearity in the return series at different stages of the modelbuilding process. Testing for chaos is a rather delicate part in this study and is done by estimating the correlation dimension for both ETFs’ return series. The correlation dimension saturates at a finite value, and the saturation indicates the presence of chaos in two ETFs considered for this study.