This paper is a preliminary report on research that is ongoing. Using unit record data from Statistics New Zealand's Income Survey for the June quarters of 1997 to 1999, we estimate wage regressions taking into account the sample selection bias problem which arises from the exclusion from such regressions of those individuals with no market income. Controlling for a set of productivity characteristics including age, household type, marital status, qualifications, occupational class and location, we find evidence of significant ethnic and gender wage differentials. In particular, we find that Maori, Pacific Island and other non-European ethnic groups do suffer labour market discrimination which is not explainable by observable characteristics. We intend to extend this study with Income Survey data from the 2000 year, incorporating useful feedback we have had since first presenting these results last year.