The main objective of this paper is to study a logarithm extension of the bimodal skew normal model introduced by Elal-Olivero et al. 1. The model can then be seen as an alternative to the log-normal model typically used for fitting positive data. We study some basic properties such as the distribution function and moments, and discuss maximum likelihood for parameter estimation. We report results of an application to a real data set related to nickel concentration in soil samples. Model fitting comparison with several alternative models indicates that the model proposed presents the best fit and so it can be quite useful in real applications for chemical data on substance concentration. Copyright © 2011 John Wiley & Sons, Ltd.