OBJECTIVE: To estimate for use in economic modeling, the time-dependent effects of deep brain stimulation (DBS) in patients with Parkinson's Disease (PD) using meta-analysis of case series. METHODS: A discrete event simulation of the course of advanced PD was created. It requires time-dependent functions of the effects of DBS. To obtain these, we searched the PUBMED, OVID and the Science Citation Index databases between 1980 and 2004 for papers reporting longitudinal experience with DBS. Data were extracted by three expert reviewers. The effect of DBS was measured at various time-points relative to baseline, while on and off medication. Time-dependent growth curves were developed by fitting the estimates as functions of time under fixed and random-effects models. RESULTS: Comparisons to baseline in the 85 studies retained showed that while off medication, activating the stimulator improved ADL rapidly (by 50.0% at 3 months) but then improvement declined slowly following a quadratic polynomial. The effect was much weaker and decline linearly while on medication but levodopa dose declined steadily, from a reduction of 590.52 (439.9-741.2) mg at 3 months to 633.8 (497.4 – 770.2) mg after 1 year. Motor skills improved by 47.2% and then more slowly following a fractional polynomial curve. CONCLUSION: These growth curves will be used to estimate the course of individual patients in simulation providing much more accurate reflection of the actual effects than traditional point estimates or transition probabilities. Given that studies can be either too small or too limited in scope to provide sound estimates of the effect of treatment, the results of meta-analytic curve fitting can be used as precise inputs to build an economic model.