Our previous work employing a between-subject randomized controlled trial design suggests that exercising prior to memory encoding is more advantageous in enhancing retrospective episodic memory function when compared to exercise occurring during or after memory encoding. The present experiment evaluates this potential temporal effect of acute exercise on memory function while employing a within-subject, counterbalanced design. In a counterbalanced order (via Latin squares), 24 participants completed four visits including (1) exercising (moderate-intensity walking) prior to memory encoding, (2) exercising during memory encoding, (3) exercising after memory encoding, and (4) a control visit (no exercise). Retrospective memory function (short term and long term; 24-hour follow-up) was assessed from a multitrial word list. Prospective memory was assessed from a time-based task. Compared to all other visits, short-term memory was greater in the visit that involved exercising prior to memory encoding (F = 3.76; P = .01; η 2 = .79). Similar results occurred for long-term memory, with no significant effects for prospective memory performance. We provide robust evidence demonstrating that acute moderate-intensity exercise prior to memory encoding is optimal in enhancing short-term and long-term memory function when compared to no exercise as well as exercising during and after memory encoding.