||Traditional age determination studies of fishes have been mainly on temperate species. Tropical species are known to be difficult to age. The validity of traditional age determination methods are evaluated in this study. Several age determination methods for Johnius belengerii (Cuvier, 1830) were carried out for samples collected from Dayawan, South China Sea. Standard length, gutted weight and some bony structures (body scales and sagittal otolith) were used in the study. Age classes were determined by length and weight frequency analysis, scale pattern reading and surface and section reading of otoliths. The results were validated by Ford-Walford method. The otolith and scale methods were equally reliable while the length frequency method was less so .The results were fitted with von Bertalanffy growth model with non-linear regression. The method resulting with the best fit was found to be otolith sections, followed by scale reading and then length/weight frequency analysis. The fit for otolith yielded the following parameters. length model: lt = 163.91 [1 - 0.9687 X e-0.8688t] weight model: wt = 125.96 [1 - 0.6167 X e-0.4258t]3 Length weight relationship was estimated. weight(g) = 7.542 X 10[to the power of negative 6] length(mm)3.199 Besides Johnius other species collected from Hong Kong fish market were also aged with the above methods, including Nemipterus virgatus (Houttuyn, 1782), Pagrus major (Temminck et Schegel, 1844), Plafycephalus indicus (Linnaeus, 1758), Konosirus punctatus (Temminck and Schegel, 1846), Sebastiscus marmoratus (Cuvier, 1829), Cynoglossus are/ (Bloch et Schneider, 1801), and Dendrophysa russelli (Cuvier, 1830). The advantages and disadvantages of these methods on different species were discussed. I would recommend the otolith section reading as the best method in age determination of tropical marine fish. Yet if the sample size is large that it is impractical to process all the otoliths, scale reading is still a good substitute. Length frequency method can be used with confidence only for very large sample sizes.