Post-weaning diarrhea (PWD) is a frequently occurring health and welfare issue in weaned piglets. Behavioral changes indicating impaired health may be detectable before the onset of signs and could be useful to detect the development of PWD early, enabling targeted and timely interventions. Current algorithms enable automated behavioral classification on the group level, while PWD may not affect all piglets in one pen and individual level analysis may be required. Therefore, this study aimed to assess whether changes in pen activity or individual piglet behavior can be early indicators of the occurrence of PWD. During 3 replicated rounds, 72 piglets (Sus scrofa domestica, Landrace x Large White) weaned at 27days of age, were housed in 4 pens with 6 piglets each. Individual fecal color and consistency were scored (0-5; ≥ 3 considered as aberrant feces) six times during the first two weeks post-weaning using rectal swabs. Additionally, using a similar scoring scale, feces on the pen floor were assessed daily. Two methods were applied for behavioral scoring. Individual behaviors (eating, drinking, standing, walking; n = 48) were scored manually and instantaneously with a five-minute interval from videos of the first two rounds, while pen activity (eating, drinking, moving; n = 12) was analyzed automatically and continuously using a commercially available algorithm from videos of all three rounds. Piglets showing a relatively higher proportion of standing behavior one day before fecal scoring had increased odds of an aberrant fecal color score (odds ratio (OR): 4.8; 95% confidence interval (CI): 1.5-15.3). Furthermore, odds of aberrant colored feces increased in pens where piglets showed more moving activity two days before (OR: 6.14; 1.26 < 95%CI < 29.84), which was also found for fecal consistency (OR: 4.77; 95%CI: 1.1-21.6). Our results indicate that increased standing in individual piglets and an increased moving activity on the pen level may be important behavioral indicators of PWD before the onset of diarrhea. Further development of current algorithms that can identify behavioral abnormalities in groups, from the pen to the individual level, may therefore be a promising avenue for improved and targeted health and welfare monitoring.