The impact of inter-system bias (ISB) on integrated precise point positioning (PPP) in multi-GNSS cannot be ignored. Precise orbit/clock products of Multi-GNSS Experiment (MGEX) satellites are diverse and gradually maturing. Analysing short-term and long-term variation characteristics of different types of ISB is thus helpful in understanding their error characteristics. In this study, ISB estimation models were proposed for GPS, GLONASS, BDS, and GALILEO systems with ionosphere-free (IF) combination observations. Then, the characteristics of ISB were mostly analyzed in detail with reverse and forward filtering method. Finally, according to the diurnal variation characteristics of ISB, which were estimated using precise products from several analysis centres, different processing strategies were adopted for ISBs to determine the optional solution strategies. Preliminary results are as follows: (1) The daily mean values of ISB estimated using the same analysis centre products between GPS and GLONASS as well as between GPS and BDS differed, but they were significantly correlated with receiver types. (2) Precise products of the same satellite system exhibited systematic deviations between analysis centres, although they were consistent on the weekly scale; however, the system bias of different satellite systems of the same analysis centres and system bias of different analysis centres all differed. (3) The daily and weekly stability of ISB estimated using different analytical centre products showed similar characteristics. Furthermore, the intra-week daily stability and weekly stability of the ISB of different stations showed good consistency. (4) Considering the strength of the observed model and the stability and reliability of the positioning results, Multi-GNSS PPP with precise products of CODE, GFZ, iGMAS and WHU showed the best performance under constant estimation, 20 min piecewise constant estimation, 1 h piecewise constant estimation and constant estimation strategies, respectively. (C) 2019 COSPAR. Published by Elsevier Ltd.