Long-term sampling is essential for monitoring the air pollutants emitted from stack since it can monitor the pollutants emission continuously including the stages of start-up, shutdown and normal operation. However, commercial continuous sampling equipment such as AMESA faces the challenges of high weight and complicated sampling procedures. This study has developed a long-term and automatic sampling system (National Central University continuous stack sampling system, NCU-CS3), and compared the efficiency with manual sampling train (MST). The results indicate that relative standard deviation (RSD) of PCDD/Fs concentrations measured between NCU-CS3 and MST is <20%, demonstrating that the difference between NCU-CS3 and MST in measuring PCDD/Fs is insignificant. Besides, the effects of adsorbent temperature, adsorbent amount and type of adsorbent on breakthroughs of PAHs and unintentional-persistent organic pollutants (UPOPs) such as polychlorinated dibenzo-p-dioxins and dibenzofurans (PCDD/Fs), polychlorinated biphenyls (PCBs), chlorinated phenols (CPs), chlorinated benzenes (CBs) and polychlorinated naphthalenes (PCNs) are evaluated. The results indicate that the breakthrough of pollutants increases with increasing temperature of XAD-2 and decreases with increasing XAD-2 amount. Moreover, XAD-4 is used as alternative adsorbent to test the breakthrough and the results indicate that the breakthroughs of UPOPs of XAD-4 as adsorbent are lower than that with XAD-2 due to higher specific surface area of XAD-4. Furthermore, the residual of PCDD/Fs with NCU-CS3 as the sampling train is relatively low (1.5–3.8%), which meets the regulation of EN 1948-5 (10%).