The purpose of this study was to develop statistical models that estimate the influence of training load on training injury and physical fitness in collision sport athletes. The incidence of training injuries was studied in 183 rugby league players over two competitive seasons. Participants were assessed for height, body mass, skinfold thickness, vertical jump, 10-m, 20-m and 40-m sprint time, agility, and estimated maximal aerobic power in the off-season, pre-season, mid-season, and end-season. Training load and injury data were summarised into pre-season, early-competition, and late-competition training phases. Individual training load, fitness, and injury data were modelled using a logistic regression model with a binomial distribution and logit link function, while team training load and injury data were modelled using a linear regression model. While physical fitness improved with training, there was no association (P = 0.16 - 0.99) between training load and changes in physical fitness during any of the training phases. However, increases in training load during the early-competition training phase decreased (P = 0.04) agility performance. A relationship (P = 0.01 - 0.04) was observed between the log of training load and odds of injury during each training phase, resulting in a 1.50 - 2.85 increase in the odds of injury for each arbitrary unit increase in training load. Furthermore, during the pre-season training phase there was a relationship (P = 0.01) between training load and injury incidence within the training load range of 155 and 590 arbitrary units. During the early and late-competition training phases, increases in training load of 175 - 620 arbitrary units and 145 - 410 arbitrary units, respectively, resulted in no further increase in injury incidence. These findings demonstrate that increases in training load, particularly during the pre-season training phase, increase the odds of injury in collision sport athletes. However, while increases in training load from 175 to 620 arbitrary units during the early-competition training phase result in no further increase in injury incidence, marked reductions in agility performances can occur. These findings suggest that reductions in training load during the early-competition training phase can reduce the odds of injury without compromising agility performances in collision sport athletes.