A critical review of the literature on security-price-changes-volume research suggests that the published studies in the United States and one each in Hong Kong and Japan have largely ignored the impacts on the results from autocorrelation, non-normality of distributions, heteroscedasticity and non-linear functional forms. Therefore, the reported findings are not robust. In testing for this relation from a small sample of continuously traded shares in the Singapore share market, we find that consistent results may not be obtained because of violations of basic test conditions. A task that remains is an application of alternative test models with data transformation using a larger sample.