OBJECTIVES: Most health economic models assessing policies for screening for diabetic nephropathy assume 100% sensitivity and specificity of screening methods. We tested the impact considering these factors on cost-effectiveness of screening for diabetic nephropathy in type 2 diabetes. METHODS: A Markov model simulated the progression of patients from no renal disease, to microalbuminuria (MAU), to gross proteinuria, and eventually to renal failure. Data were derived from published sources. Costs and life expectancy (LE) (discounted at 3% p.a.) and incremental costs/life year gained (C/LYG) were calculated for an MAU screening program (treating with ACE inhibitors if detected) versus no screening assuming 100% sensitivity and specificity of the screening method (scenario 1), and were compared to those calculated with more realistic values of sensitivity 71% and specificity 98% (scenario 2). A US Medicare perspective was taken. RESULTS: Costs and LE in a typical type 2 population for scenario 1 were: $7047 and 12.70 years in the screened population, and $11465 and 12.43 years in the unscreened population, with screening reducing overall costs and increasing life expectancy. For scenario 2, costs and LE in the screened population were $7303 and 12.66 years, (and remained unchanged in the unscreened population). Sensitivity analysis of the sensitivity of screening showed that at sensitivities below 50%, LE and costs exponentially approached those of the no screening strategy. CONCLUSIONS: While affecting the absolute values for costs and LE, more realist assumptions about the sensitivity and specificity of screening methods for nephropathy had little impact on the relative results, with a nephropathy screening program dominant to no screening under both sets of assumption. If sensitivity of a nephropathy screening test is <50%, sensitivity should be incorporated in future nephropathy health economic models assessing screening intervention policies.