Strength prediction by non-destructive test techniques always used statistical tools called regression technique. The reliability of this technique is always judged based on the value of coefficient of determination (r 2) or correlation coefficient (r) of that equation. This technique alone is not always reliable. Thus, there is a need to compliment it with some other statistical techniques. Hypothesis test seems to be one of them, which can predict the reliability of those equations obtained from regression analysis. Four different mix proportions in two series namely, A10 and A 19 incorporating different aggregate (sandstone) sizes (10 mm and 19 mm maximum) were investigated to predict correlations between strength and Sclerometer test of concrete. Regression models and their coefficient of determination were obtained for each mix proportions, each series and the combined series. The coefficient of correlation value was found to be 0.97 for overall regression model and this articulates reliability of the model. The hypothesis tests on the regression model confirm their reliability. © 2008 Taylor & Francis Group.