Following the earlier analysis, this paper attempts to explain the variations in unit costs and service levels across the sample villages in Andhra Pradesh. The analysis is based on the data collected at the village and household level. Multiple regression analysis is based on the data collected at the village and household level. Multiple regression analysis is used to estimate the cost drivers and factors influencing service levels. Number of dependent and independent variables have been identified from the available household data. The selected specifications are fairly good in explaining the variations in unit costs and service levels and the coefficients are consistent. The analysis of drinking water suggests that the relation between existing unit costs focused almost exclusively on capital expenditure and service levels is quite weak. Level of education, demographic and economic factors, technology and governance related factors appear to be influencing unit costs as well as service levels. The analysis not only emphasises some of the earlier observations but also provides new insights into explaining the variations.