When treatment complications are the main criteria of a study, the simplest way for estimating the complication rate over time is to use the Kaplan-Meier method. The failure is the onset of the first complication. A new way is proposed here which takes into account the time to an eventual second, and third complication, and their duration as well. Results from a phase III randomized trial comparing two preoperative brachytherapy low dose rates in localized stage cervix carcinoma patients are used to illustrate both methods. The new method showed a greater long term complication rate in the higher dose rate group.