With a view to see the empirical validity of random walk hypothesis in India with the 'variance ratio test', the present paper has made analysis at two levels i.e at aggregated and disaggregated levels. While the disaggregated level made use of daily closing prices of a sample of individual securities, the aggregated level made use of the closing price series of four popular stock market indices of national stock exchange for the period 2001 to 2007. The results of aggregate level analysis indicate the existence of non-randomness in the movements of Nifty. Contrary to this, the results of disaggregate level analysis are found to be supporting the existence of random walk hypothesis when the test statistic is adjusted for hetroscedasticity. It indicates that the appearance of stock price dependency on the past price data in Indian bourses is not real as claimed by a section of analysts, but because of the spurious correlation and presence of hetroscedasticity. Nifty being the most popular stock market index in India, it is expected to reflect the overall changes in the economy hence the non-randomness in its movements is not surprising.