In most heat transfer applications, knowledge of the viscosity variation is important. Thus viscosity correction factors have been researched and proposed for almost a century. One of the most successful relations was reported by Sieder-Tate in 1936, which has been widely used in engineering analysis and design. In this study, we have improved on the Sieder-Tate relation, following a classical theoretical analysis of the thermal boundary layer. An exact solution to the viscosity correction factor was obtained which shows that the Sieder-Tate correction factor over-predicts the heat transfer coefficient in the case of cold wall (cooling) and does not hold properly for hot wall (heating). We have found that a relation of (mu(infinity)/mu(w))(0.254) (in case of cooling) and (mu(infinity)/mu(w))(0.087) in the case of heating is better than the Sieder-Tate factor of (mu(infinity)/mu(w))(0.14) (uniformly applied to both the heating and cooling cases). Here mu(infinity) and mu(w) are the bulk and wall viscosity coefficients respectively. The theoretical analysis also shows that the above correction factors are limited to small values of ln (mu(w)/mu(infinity)) (for cold wall) and ln (mu(infinity)/mu(w)) (for hot wall). However a general solution has been obtained and the correlation developed by Petukhov (1970) closely matches the exact solution for the case of cold wall cooling of a fluid. (C) 2018 Elsevier Ltd. All rights reserved.