This study replicates and extends Dusenbury (1994). A sample of 132 New Zealand taxpayerparticipants performed a laboratory experiment consisting of four decision problems relating to tax situations and two decision problems relating to other financial contexts. The focus of interest was the power of each tax context's stated prepayment position to induce a decision frame (as predicted by Prospect Theory) capable of determining each participant's risk preferences. The use of a repeated measures analysis of variance procedure (general linear model) allowed for within-subject comparisons so that decision-frame-induced shifts in individual participants' risk preferences across decision problems could be diagnosed. The study found, with respect to the tax problems, that the predictions of Prospect Theory and the findings of Dusenbury (1994) were generally supported. However, similar risk preference shifts based on alternative decision frames, independent of prepayment position and all other contextual information, were also detected. When within-subject comparisons were also generated by contrasting of tax and non-tax decision problems, identical apart from context, it was found that participants produced evidence of dissimilar, context-sensitive risk preferences. Risk preferences were also found to be stable for decision problems containing identical option sets and similar, but not identical contexts. Also, a significant behavioural difference was detected between participants with high cash floats and those with low cash floats; but this was not differentiated from an awareness of cash float status variable in the study. Finally, while the underlying value functions of the participants conformed to the predictions of Prospect Theory in the tax contexts, there were some departures from the predicted form when the decision problem was set in a gambling context.