The clinical efficacy of roflumilast, an oral phosphodiesterase-4 inhibitor, has been demonstrated in patients with severe chronic obstructive pulmonary disease (COPD). However, roflumilast has shown frequent adverse drug reactions (ADRs). This study was performed to investigate the dosing strategy that will improve adherence to roflumilast in COPD. We conducted a systematic review and meta-analysis using PubMed, Embase, and Cochrane Central Register. The dosing strategy for roflumilast was classified into a dose-escalation group and a low-dose group. We investigated clinical outcomes according to dosing strategy. Five clinical trials involving 2424 patients were included. Both the dose-escalation and the low-dose groups showed a decrease in discontinuation rate compared to the standard dosing group for roflumilast (risk ratio [RR], 0.81; 95% confidence interval [CI], 0.67-0.97; P = 0.02 and RR, 0.62; 95% CI, 0.48-0.80; P < 0.01, respectively). In the two strategies, the pooled proportions of discontinuation were 27.9% and 11.7%, respectively. Although the pooled proportion of any ADR was not statistically decreased in the two strategies, diarrhea was significantly reduced in the low-dose group compared to the standard group (RR, 0.58; 95% CI, 0.42-0.82; P < 0.01). The pooled incidence of acute exacerbations was similar between the low-dose and the standard groups (22.9% and 20.1%, respectively; P = 0.27). Our findings show that the two alternative dosing strategies might have the benefit of improving adherence to roflumilast in COPD. Further large-scale trials are required to support our findings.