Introduction & Purpose The Global Physical Activity Questionnaire (GPAQ; Armstrong & Bull, 2006) is commonly used to assess frequency, duration and intensity of physical activity (PA) in three domains: work, travel and recreation. It is frequently used as a self-administered questionnaire; however, the execution was originally designed for an interviewer. With the continuous development of app-based technology, the evaluation of PA online-assessment is relevant (Mosa et al., 2012). The objective of the study was to compare an online self-administered version of the GPAQ with the original version administered by an interviewer. Methods The physical activity behaviour of 533 participants (70% female, average age 44, ± 13.9) from the LIFE-Tirol Study (Lasting Health through Intermittent Fasting, Emotional Firmness and Exercise) was assessed by an interviewer using the GPAQ. The interviewer used generic show cards for explaining different levels of exercise intensities as recommended in the analysis guidelines (Armstrong & Bull, 2006). After two months, participants were asked to fill in the GPAQ via online app without show cards. Participant were advised not to change their usual physical activity levels during the two months. The movement domains and intensities of the two versions were compared with Jamovi v2.3 using a series of Wilcoxon W-tests and a series of 2 × 2 mixed ANOVAs to analyse gender- and age-specific aspects. Results Total metabolic equivalent task (MET) minutes were significantly different between interview and self-assessed online version (1,971 ± 2,590 vs. 4,143 ± 4,370 MET minutes; p < .001, d = 0.670). Significantly more MET minutes were reported in the online version compared to the interview version across all three PA-domains and both vigorous and moderate intensities (F(1, 552) = 167, p < .001, ƞ² = .232). This difference was not influenced by gender. 18–34-year-old participants reported significantly less MET minutes during recreational PA than participants older than 55 years in both versions (F(1, 551) = 113, p < .001, ƞ²= .170). No further gender/age × assessment interaction effects were found. Male participants tend to report higher intensity minutes more likely than females in both assessment versions (F(1, 551) = 4.43, p = 0.360, ƞ²= .008). Discussion The amount of PA as well as its intensities seem to differ between the self-assessed online version compared to the interview assessment of the GPAQ, although the interview was conducted as the first measurement and PA as well as intensities were explained in detail to participants using generic show cards. It was previously reported that show cards have no significant impact on the results of the self-administered version of the GPAQ (Rudolf et al., 2020). Therefore, the authors suggest that the main differences between both assessments were caused by the presence of the interviewer independent of age or gender. Conclusion Regardless of the use of show cards and detailed explanations of PA, the main difference between guided and self-administered PA-assessments may result from the presence or absence of an interviewer. Therefore, it is important to interpret self-administered app-based GPAQ evaluations with caution. Future studies should validate app-based GPAQ data by using objective PA measurements, such as accelerometers.