From previous systematic reviews and meta-analyses, there is consensus about the positive effect of exercise training on exercise capacity for systolic heart failure (HF); however, the effect on actual prognostic markers such as NTproBNP and minute ventilation/carbon dioxide production (VE/VCO2) slope has not been evaluated. The primary aim of the proposed study is to determine the effect of aerobic exercise training (AEX) on the VE/VCO2 slope and NTproBNP. The following databases (up to February 30, 2013) were searched with no language limitations: CENTRAL (The Cochrane Library 2013, issue 2), MEDLINE (from January 1966), EMBASE (from January 1980), and Physiotherapy Evidence Database (PEDro) (from January 1929). We screened reference lists of articles and also conducted an extensive hand search of the literature. Randomized controlled trials of exercise-based interventions with 2-month follow-up or longer compared to usual medical care or placebo were included. The study population comprised adults aged between 18 and 65 years, with evidence of chronic systolic heart failure (LVEF < 45 % and baseline NTproBNP > 300 pg/ml). Two review authors independently extracted data on study design, participants, interventions, and outcomes. We assessed the risk of bias using PEDro scale. We calculated mean differences (MD) or standardized mean differences between intervention and control groups for outcomes with sufficient data; for other outcomes, we described findings from individual studies. Eight studies involving a total of 408 participants met the inclusion criteria across the NTproBNP (5 studies with 191 patients) and VE/VCO2 slope (4 studies with 217 patients). Aerobic exercise significantly improved NTproBNP by a MD of -817.75 [95 % confidence interval (CI) -929.31 to -706.19]. Mean differences across VE/VCO2 slope were -6.55 (95 % CI -7.24 to -5.87). Those patients' characteristics and exercise were similar (frequency = 3-5 times/week; duration = 20-50 min/day; intensity = 60-80 % of VO2 peak) on the included studies. Moreover, the risk of bias across all studies was homogeneous (PEDro scale = 7-8 points). However, based on the statistical analysis, the heterogeneity among the studies was still high, which is related to the variable characteristics of the studies. Aerobic exercise may be effective at improving NTproBNP and the VE/VCO2 slope in systolic HF patients, but these effects are limited to a specific HF population meeting specific inclusion criterion in a limited number of studies. Future randomized controlled studies including diastolic and HF overleap with pulmonary diseases are needed to better understand the exact influence of AEX.