Public transport is a potentially important part of independent living for older people, but they are over-represented in non-collision bus injuries. This paper reports on a computational modelling approach to addressing this problem: the Madymo human model validated for simulating passive, seated vehicle occupants was adapted to simulate a standing passenger in an accelerating bus. The force/deformation characteristics of the bus were measured and the human model was expanded to include a validated active hand grip. Real world urban bus acceleration profiles were measured and used as inputs for the simulations. Balance loss could not be predicted, but injuries from contact with the vehicle floor following a fall were evaluated. Results show that peak bus accelerations measured (+/-0.32g) exceed reported acceleration thresholds for balance loss for a standing passenger using a handgrip (0.15g). The maximum predicted probability of knee and head injuries arising from impact with bus seats, handrails and walls were 53% and 35%, respectively. The stairwell and horizontal seatback handles were particularly hazardous and the latter should be replaced with vertical handrails. Driver training should be expanded to include video training based on multibody simulations to highlight the risks for standing passengers induced by harsh braking and acceleration.