Assessment of neurocognitive function (NCF) is important in brain tumor clinical trials, however there are varying methodologies available. We used the Cogstate computerized NCF testing battery and the mini-mental state examination (MMSE) to prospectively assess cognition in adult patients with recurrent glioblastoma (GBM) enrolled in the CABARET randomized phase II clinical trial of bevacizumab versus bevacizumab plus carboplatin chemotherapy. We determined completion rates; compared NCF results between trial arms; and assessed baseline NCF as a predictor of survival outcome. 93 of 103 eligible patients completed baseline Cogstate NCF testing. Completion rates were between 60 and 100% across each timepoint, and 38% at disease progression. There was no evidence of difference between arms in time to deterioration in NCF using either test. Prior to disease progression, deterioration on the Cogstate tests was substantially more common (90%) than deterioration on the MMSE (37%), and decline in the Cogstate composite score within the first 8 weeks was associated with shorter overall survival. This testing methodology may be useful when determining net clinical benefit for therapies in patients with recurrent GBM.