this study used the theoretical framework of pender's health promotion model (hpm) to predict physical activity patterns in a sample of spinal cord injured (sci). the model was adapted to better fit the population under study by including coping style, level of lesion and weight as individual characteristics and physical self-efficacy as a behaviour-specific cognition. the relationships of personal factors, coping strategies, social support, barriers to action, and physical self-efficacy were explored. a sample of 33 respondents who had either quadriplegia (19) or paraplegia (14) completed the study instruments. there were 29 males and 4 females with a mean age of 40.5 years and a mean age of injury of 24 years. participants completed each of six instruments including; physical self-efficacy (pse) and its two subscales perceived physical ability (ppa) and physical self-presentation confidence (pspc); exercise benefits barrier scale (ebbs); jalowiec coping scale (jcs); personal resource questionnaire (prq- 85); leisure time exercise questionnaire (lteq); and a personal inventory for demographic and health risk information. it was hypothesized that high physical self efficacy, few barriers to action, and high social support would be positively related to involvement in physical activity, and conversely, that high lesion levels (quadriplegia), overweight and use of an affective coping style would relate to low involvement in physical activity. results of the present study provide some support for applying the adapted model of health promotion to explain the physical activity patterns of people with sci. participation in moderate, strenuous, sweat inducing and total weekly physical activity was moderately correlated to high levels of both ppa and pse. a clear but weaker relationship was also shown between few perceived barriers to participation and involvement in mild, strenuous and total weekly physical activity. in addition, severity of impairment or comparison of activity level to lesion level indicated that paraplegics were more involved in strenuous activity than quadriplegics. no relationship was found between level of social support or affective copying style and physical activity. the strongest predictors of physical activity were high ppa and short time since injury (tsi). sixty-seven percent of the variance in participation in strenuous activity was accounted for by ppa (38%), height (15%) and age (14%). fifty-five percent of the variance in total weekly activity (lteq) was explained by ppa (35%), tsi (11%) and stiffness and soreness (9%). fifty-one percent of the variance in mild activity was explained by tsi (36%) and pspc (15%) and 48% of moderate exercise was explained by tsi (34%) and ppa (14%). as a consequence of this investigation, a revised model of health promotion was presented. it was concluded that those people with sci who adopt an active approach to health promotion are young, recently injured, have a high ppa and are paraplegic. it was recommended that rehabilitation practices concentrate on increasing physical self efficacy and empowerment through the use of situation-specific learning environments and a variety of activities that have a life span approach.