The aim of this study was to investigate whether direct PCR (DP) gave similar results to culture prior to PCR (CPP) for detecting mycoplasmas in different types of pig tissues. A total of 724 samples obtained from lungs, tonsils, or synovial fluids from 270 slaughtered pigs were used. The history of clinical signs, lung score, and the presence of joint lesions were recorded during sample collection. The rates of detection of Mycoplasma hyopneumoniae, Mycoplasma hyosynoviae, and Mycoplasma hyorhinis using both procedures were evaluated. The overall prevalences of M. hyopneumoniae, M. hyosynoviae, and M. hyorhinis were 40.3%, 12.3%, and 64.6%, respectively, and the detection rate depended on the sample type and the procedure used. With lung tissue, DP gave a higher detection rate for M. hyopneumoniae (77.4%) than CPP (38.5%). M. hyorhinis was detected by CPP at 15.6% and 18.1% and by DP at 31.5% and 5.2%, respectively. The positive rate derived from tonsil from CPP was closed to that of DP. Using synovial fluid could not yield any positive M. hyorhinis from CPP whereas 37.2% was positive from DP. In contrast, using sample tissue from lung and tonsil by CPP could show much higher positive number than that of DP. There was a significant relationship between joint lesion and M. hyorhinis detection by DP (P < 0.05) but not for M. hyosynoviae and M. hyorhinis detected by CPP. We speculated that lung was a proper sample for M. hyopneumoniae and M. hyorhinis detection by DP and CPP, respectively. Tonsil was likely the community of persistent M. hyosynoviae and M. hyorhinis with highly detection by CPP. Synovial fluid was apparently unsuitable for mycoplasmal culture. The accuracy of mycoplasmal detection may depend upon the type of sample relevant to the detection procedure used.