This study investigated the effects of drink composition on voluntary intake, hydration status, selected physiological responses and affective states during simulated gymnasium-based exercise. In a randomised counterbalanced design, 12 physically active adults performed three 20-min intervals of cardiovascular exercise at 75% heart rate maximum, one 20-min period of resistance exercise and 20 min of recovery with ad libitum access to water (W), a carbohydrate-electrolyte solution (CES) or with no access to fluids (NF). Fluid intake was greater with CES than W (1706 ± 157 vs. 1171 ± 152 mL; P < 0.01) and more adequate hydration was achieved in CES trials (NF vs. W vs. CES: −1668 ± 73 vs. −700 ± 99 vs. −273 ± 78 g; P < 0.01). Plasma glucose concentrations were highest with CES (CES vs. NF vs. W: 4.26 ± 0.12 vs. 4.06 ± 0.08 vs. 3.97 ± 0.10 mmol/L; P < 0.05). Pleasure ratings were better maintained with ad libitum intake of CES (CES vs. NF vs. W: 2.72 ± 0.23 vs. 1.09 ± 0.20 vs. 1.74 ± 0.33; P < 0.01). Under conditions of voluntary drinking, CES resulted in more adequate hydration and a better maintenance of affective states than W or NF during gymnasium-based exercise.