OBJECTIVES: In probabilistic sensitivity analysis (PSA), it is typical to see distributions assigned to all (relevant) parameters in a model. However, attention is only usually paid to estimating covariance or interactions between a small number of parameters, if any at all. The study explores the impact of interaction and non-interaction assumptions on the outcomes of PSA. METHODS: A range of very simple models (additive, multiplicative and ratio-based) were developed, with corresponding input parameters. For each model, a range of alternative approaches were taken. These included adding 'sub-level' inputs to each parameter to add detail (for example, rather than a single input parameter for 'monthly cost of health state X', individual parameters were created for 'cost of physician visits', 'cost of tests', 'cost of drugs', 'cost of hospital visits', etc. These were all varied independently in the PSA. RESULTS: The implications of ignoring parameter interactions in PSA varied widely depending on the type of model. Models where QALYs and costs are likely to be positivity correlated (i.e. in survival-based models such as oncology) displayed the opposite effect to those where QALYs and costs are inversely related (i.e. for long-term chronic conditions such as diabetes). Furthermore, the analysis demonstrates that, if a specific input parameter is broken down into several components which are varied independently, then it is likely that the variation in each parameter will cancel out the effect of the changes in the other parameters. Indeed, the greater the 'granularity' of the input parameters, the greater the likelihood that the variations will offset each other, this suggesting a false level of certainty in the PSA's results. CONCLUSIONS: This analysis demonstrates the outcomes of a PSA can be influenced by the level of detail that the modellers choose to include and, counterintuitively, modellers can create 'false' confidence in PSA results by including more parameters.