We sought to determine whether short-term neoadjuvant androgen deprivation (STAD) duration influences the optimal time point from which Phoenix fail (time to biochemical failure; TTBF) should be measured. In the Trans-Tasman Radiation Oncology Group 96.01 trial, men with locally advanced prostate cancer were randomized to 3 or 6 months STAD before and during prostatic irradiation (XRT) or to XRT alone. The prognostic value of TTBF measured from the end of radiation (ERT) and randomization were compared using Cox models. Between 1996 and 2000, 802 eligible patients were randomized. In 436 men with Phoenix failure, TTBF measured from randomization was a powerful predictor of prostate cancer-specific survival and marginally more accurate than TTBF measured from ERT in Cox models. Insufficient data were available to confirm that TTBF measured from testosterone recovery may also be a suitable option. TTBF measured from randomization (commencement of therapy) performed well in this trial dataset and will be a convenient option if this finding holds in other datasets that include long-term androgen deprivation data.