OBJECTIVES: To compare and evaluate freeware options for bootstrap analyses of incremental cost effectiveness data. METHODS: Obenchain's ICEplane software can be downloaded and installed from www.math.iupui.edu/~indyasa/bobdown.htm. This software was compared against the web-based bootstrap analyses available online at HealthStrategy.com. Three datasets, that are downloadable from the ICEplane site, were used in this software comparison. The three datasets are from published studies dealing with abciximab (ABX), pindolol (PIN) and tricyclic antidepressants (TCA). RESULTS: ICEplane must be installed on MS Windows operating systems, in contrast to the HealthStrategy program that runs online through any operating system with browsers including Internet Explorer, Netscape or Firefox. Both software options provide output graphs such as scatter plots, confidence intervals and acceptability curves. ICEplane and HealthStrategy generated very similar statistics on the initial raw data such as mean, median, standard deviation and standard error. Bootstrapped statistics include mean, median, incremental cost effectiveness ratio (ICER), and 95% confidence intervals. Respective mean ICERs and confidence intervals between ICEplane and HealthStrategy bootstraps on the three datasets were as follows: ABX: 3,771 (-665, 20,797) vs. 3,692 (-1,054, 13,303); PIN: -1,880 (-412, 2,863) vs. -1,832 (-333, 25,442) and TCA: -16.48 (-169, 136) vs -19.96 (-221, 147). There was good agreement on Fieller's confidence intervals. CONCLUSIONS: The ICEplane software has more statistical and charting features than the HealthStrategy bootstrap program. The analyses from HealthStrategy ran more slowly with over 1000 bootstrap replications, but the results obtained compare reasonably well to ICEplane. The HealthStrategy site has the potential benefits of requiring no installation and accessibility on multiple computer platforms. Both of these freeware options should make it easier for individuals to explore basic bootstrap analyses of cost effectiveness data, but more comprehensive statistical packages like Stata should be used when possible.