Limit states design methods and engineering judgement have been used to assess buoyancy issues for remediation of the 85 year old Shoal Lake Aqueduct in Manitoba. The study demonstrates how these methods can be applied to non-codified engineering problems. Four separate buoyancy analyses were completed using (i) partial safety factors from the Ontario Highway Bridge Design Code, (ii) project-specific partial safety factors, (iii) Monte Carlo simulation techniques, and (iv) working stress design (WSD) methods. Engineering judgement was required to develop a buoyancy model, interpret data for modeling parameters, and provide meaningful values for parameters that could not be measured. Results from the analyses show that more uniform reliability is provided when measured variability of the model parameters is accounted for. The reliability is not quantifiable when working stress design methods are used.