Injury prevention efforts have been hindered by a lack of effective surveillance in agriculture. The objective of this cohort study was to use a school-based tracking system for migrant farmworker students as a surveillance method to identify migrant farmworker families in Starr County, located along the Texas-Mexico border, in order to estimate the frequency and risk factors for occupational injuries over a two-year period. Injuries were defined as getting hurt when working or traveling to or from work while migrating in the past year. Rio Grande City Consolidated Independent School District (RGCCISD) is the largest of three school districts in Starr County, and enrolled nearly 60% of all students in the county. The Migrant Education Program (MEP) in RGCCISD maintained files computerized through the New Generation System (NGS) on 1,200 families in the 1998-99 school year. Randomly sampled from the New Generation System, our cohort consisted of 267 families, who in the spring of 2000 responded to a screening questionnaire and indicated their intention to migrate. All were Hispanic. The average family size was 4.7 people. Mothers and fathers had an average of 7 years of education and had lived in Starr County for an average of 17 years (mean for mothers). The mother was the primary respondent for the family, and questions were asked about the mother, father, oldest and youngest child (Year 1 follow-up only). Approximately, 57.7% in Year 1 and 59.8% in Year 2 of eligible participant families migrated (n=154 at Year 1 follow-up and 143 at Year 2 follow-up) and 96% of these completed the follow-up interviews. These families represented about 310 individuals each year who had worked in the fields on average 6 days a week, 10 hours a day, for 2.7 months in the past year. Twenty-five work-related injuries (first injury events only) were reported during the 2 years of follow-up. The overall rate of injury was 12.52/100 FTE (95% C.I., 8.55-19.01). The most frequent injuries were cuts/jabs, transportation-related injuries (including head injury), blisters/rashes from pesticides, and heat exhaustion/stroke. Over one-third of the injuries resulted in lost work time. On average each year, 30% of mothers, 22% of fathers, 15% of oldest children and 10% of youngest children (Year 1 follow-up only) also reported chronic back pain. In terms of hazard surveillance, from 1/3 to 2/3 of all family members who worked in the fields were exposed to tractors, knives, chemicals, repetitive lifting (fathers). Of 102 mothers who participated in migrant farm work during the summer of 2001, only 44 (42%) reported having received EPA regulated pesticide safety training within the previous 5 years. Cox regression was used to examine risk factors for first injury events. Significant predictors of injury included employer-type, working with or around chemicals, general use of motor vehicle seat belts, number of farm jobs during the migration season, and employer-providing toilet paper. Logistic regression was used to examine chronic back pain and hand symptoms experienced during the last migration season. Factors associated with chronic back pain included age, depression during the last migration season, repetitive bending/stooping, hours of sleep while migrating, hours of sleep at home in Starr County, self-provided drinking water, sorting as a job task, and exposure to pesticide drift/spray. Factors associated with hand problems included age, working with or around chemicals, repetitive handwork, and average hours of work per day. In addition, two methodological sub-studies were completed. A work diary (calendar with number of hours worked, tasks, crops and injuries/illnesses) was developed and the feasibility tested. Fifty mothers were invited to participate in the piloting of a work diary. Of invited mothers (n=50), 30 mothers migrated and were eligible to participate. Of these, 10 mothers returned a least some portion of the work diary (response rate, 33.3%). In addition, surveillance methodologic issues were examined regarding the feasibility/validity of asking the mother to respond on behalf of her family members. Concordance between mothers and a sub-sample of the fathers and children was assessed by computing kappa statistics and sensitivity/specificity for a subset of variables including acute injury, work hazards, and illness symptoms. Regarding reporting acute injury, kappa statistics for the father/mother and child/mother pairs were 0.84 (almost perfect) and 0.45 (moderate), respectively. Using the same pair order, the sensitivities were 0.75 and 0.40, and the specificities were 1.00 and 0.97, respectively. However, for back pain, the kappa statistics were reduced and back pain frequency as reported by the mother for both father and child was underestimated. The New Generation System proved to be an effective method to identify migrant farmworker families in Starr County for epidemiologic studies or future interventions. Given that a majority (58%) of the migrant farmworker women in our study had not received Worker Protection Standard employer-mandated pesticide safety training, increased enforcement, education, and alternative delivery of this training are recommended. If person-time at risk for injuries is taken into account as well as symptoms of chronic back pain, the reported injuries are substantial. These data support the need for primary data collection from young farmworkers, systematic streamlined surveillance, and school-based and community interventions.