Background The effects of combined training can be affected by training characteristics such as frequency, session length, and duration. No empirical studies to date have directly compared how combined physical and cognitive training offered at different training frequencies affects cognitive function for older adults with cognitive decline. This study investigated the impact of training frequency on cognitive outcomes after combined physical and cognitive training for older adults with cognitive decline. Methods A quasi-experimental study was conducted in community facilities and day care centers. The study assigned 89 older adults with cognitive decline into high-frequency (HF) or low-frequency (LF) training groups. The participants received 90- to 105-min training sessions, one (LF) or three (HF) times a week, for 12 weeks. Outcome measures were the Montreal Cognitive Assessment, Word List subtest of the Wechsler Memory Scale, Digit Symbol Substitution Test (DSST), and Stroop Color Word Test. Results The HF group demonstrated greater improvement in immediate memory measured by the WL-IM (F = 8.7, P = 0.004) and in executive function measured by the SCWT (F = 5.89, P = 0.017) than the LF group. Compared with the HF group, the LF group showed a great improvement in delayed memory measured by the WL-DM (F = 9.62, P = 0.003). The HF and LF groups both increased in processing speed and global cognitive function. Conclusions Our study indicated that the different training frequency of combined physical and cognitive training may result in benefits on different cognitive functions in older adults with cognitive decline. These findings may assist clinical practitioners in choosing appropriate training frequencies based on various intervention purposes for the elderly with cognitive decline. Trial registration ClinicalTrials.gov Identifier NCT03619577 (08/08/2018).