The purpose this study was to examine the effects of caffeine ingestion on performance and energy expenditure (anaerobic and aerobic contribution) during a 4-km cycling time trial (TT) performed after a carbohydrate (CHO) availability-lowering exercise protocol. After preliminary and familiarization trials, seven amateur cyclists performed three 4-km cycling TT in a double-blind, randomized and crossover design. The trials were performed either after no previous exercise (CON), or after a CHO availability-lowering exercise protocol (DEP) performed in the previous evening, followed by either placebo (DEP-PLA) or 5 mg.kg(-1) of caffeine intake (DEP-CAF) 1 hour before the trial. Performance was reduced (-2.1%) in DEP-PLA vs CON (421.0±12.3 vs 412.4±9.7 s). However, performance was restored in DEP-CAF (404.6±17.1 s) compared with DEP-PLA, while no differences were found between DEP-CAF and CON. The anaerobic contribution was increased in DEP-CAF compared with both DEP-PLA and CON (67.4±14.91, 47. 3±14.6 and 55.3±14.0 W, respectively), and this was more pronounced in the first 3 km of the trial. Similarly, total anaerobic work was higher in DEP-CAF than in the other conditions. The integrated electromyographic activity, plasma lactate concentration, oxygen uptake, aerobic contribution and total aerobic work were not different between the conditions. The reduction in performance associated with low CHO availability is reversed with caffeine ingestion due to a higher anaerobic contribution, suggesting that caffeine could access an anaerobic "reserve" that is not used under normal conditions.