In this paper we present a class of ratio type estimators of the population mean and ratio in a finite population sample surveys with without replacement simple random sampling design, where information on an auxiliary variate x positively correlated with the main variate y is available. Large sample approximations to mean square errors (MSE) of these estimatorsare evaluated and their MSE's are compared with the MSE of the usual ratio estimator _R of the population mean of y. It is shown that under certain conditions these estimators are more efficient than _R. When a prior knowledge of the value of thecoefficient of variation, c_y, of y is at hand, ratio type estimator, say ₁ of is proposed. It is shown, under certain conditions, that ₁ is more efficient than _R. When values of c_y, c_x and the population correlation coefficient ρ is at hand, then we have proposed another estimator, say ₂ of , which is always better than _R as far as the efficiency is concerned. In fact, is ₂ is shown to be even better than ₁. Finally estimators better than the usual ratio estimator / of are given.