The American Association of Physicists in Medicine Task Group 119 instructed institutions to use a low‐dose threshold of 10% or a region of interest determined by the jaw setting when they collected gamma analysis quality assurance (QA) data for the planar dose distribution. However, there are no clinical data to quantitatively demonstrate the impact of the low‐dose threshold on the gamma index. Therefore, we performed a gamma analysis with various low‐dose thresholds in the range of 0% to 15% according to both global and local normalization and different acceptance criteria (3%/3 mm, 2%/2 mm, and 1%/1 mm). A total of 30 treatment plans — 10 head and neck, 10 brain, and 10 prostate cancer cases — were randomly selected from the Varian Eclipse treatment planning system (TPS). For the gamma analysis, a calculated portal image was acquired through a portal dose calculation algorithm in the Eclipse TPS, and a measured portal image was obtained using an electronic portal‐imaging device. Then, the gamma analysis was performed using the Portal Dosimetry software (Varian Medical Systems, Palo Alto, CA). The gamma passing rate (%GP) for the global normalization decreased as the low‐dose threshold increased, and all low‐dose thresholds led to %GP values above 95% for both the 3%/3 mm and 2%/2 mm criteria. However, for the local normalization, %GP for a low‐dose threshold of 10% was 7.47%, 10.23%, and 6.71% greater than the low‐dose threshold of 0% for head and neck, brain, and prostate for the 3%/3 mm criteria, respectively. The results indicate that applying the low‐dose threshold to global normalization does not have a critical impact on patient‐specific QA results. However, in the local normalization, the low‐dose threshold level should be carefully selected because the excluded low‐dose points could cause the average %GP to increase rapidly.PACS number: 87.55.Qr