In this paper, built-in sensors were described to automatically detect human daily activities. In contrast to the previous work, this paper intends to recognize the physical activities when the phone’s orientation and position are varying. The data collected from six positions of seven subjects were investigated and two signals that are insensitive to orientation were chosen for classification. Decision trees (J48), Naive Bayes and sequential minimal optimization (SMO) were employed to recognize five activities: static, walking, running, walking upstairs and walking downstairs. The classification results of three classifiers were compared. The results demonstrates that the J48 classifier produces the best performance (average recognition accuracy: 90.7%). Then we chose the J48 classifier as online classifier.