OBJECTIVE： To provide scientific reference for prescription comment. METHODS： Network questionnaire was designed to investigate the development of hospital prescription comment taking hospital pharmacists from Shanghai as the object. The surveyed data were analyzed statistically. RESULTS： A total of 51 effective questionnaires were collected with effective recovery rate of 100％， involving 51 hospitals. Second-level prescription comment mode was adopted in 51 hospitals； the mode of “initial clinical pharmacist comment， expert group re-comment” was mostly adopted in second-level hospitals， accounting for 23.53％. The mode of “outpatient pharmacist initial comment， clinical pharmacist re-comment” was mostly adopted in third-level hospitals and general hospitals， accounting for 23.53％ and 21.57％， respectively. The method of “all prescriptions randomly selected at one stage” was mostly adopted in third-level hospital （accounting 23.53％）. 45.10％ of them did not adopt information system for prescription comment. 86.27％ of surveyed hospitals carried out the review once a month. The limitation on the amount of drugs which were “less than 2 weeks” and “less than 1 month” took up the high proportion （41.18％， 45.10％）. Among irrational used common drugs， antibiotics accounted for the highest proportion （43.14％）. In supernormal prescriptions， the most common problem was drug use without indication （accounting for 64.71％）. Surveyed pharmacists from different types of hospital had different comments on same prescription. The effects of prescription comment on hospital rational drug use and pharmacist’s ability had been widely recognized， but its effects on physician prescribing behavior was not satisfactory （only 5.88％ respondents believed the great effects of prescription comment）. Main ideas and suggestions of surveyed pharmacists for improving the role of prescription comment were “to improve the comment ability of pharmacists”， “to pay more attention by hospital leaders” and “to increase the degree of information management”. CONCLUSIONS： During the process of prescription comment， it is necessary to improve the authority of prescription comment and the scientific sampling method， optimize the frequency of prescription comment and the consistency of comment results， and also pay attention to combination with administrative intervention.