In this paper, we tested three types of asymmetry of business cycles: steepness, deepness and sharpness, using the parametric tests of Clements and Krolzig (2003), based on Markov-Switching autoregressive models in the real GDP of Malaysia for the period from 1975Q1 to 2006Q4. This study used three different approaches to remove trend component of GDP to investigate the sensitivity of the findings of asymmetries to the method of trend eliminations. Finally, for comparison, the non-parametric test of Randles et al. (1980) was also employed. We found a strong evidence of steep asymmetry but not deep sharp asymmetry for Malaysia which is in contrast to the findings of Khong and Lau (2007) and Eng and Wang (2008).