Summary Slipping on ice is one of the main risks of outdoor occupational activities during winter. Ice and freezing rain were involved in 14% of slip, trip and fall (STF) accidents from 2014 to 2016 in Quebec. Many workers who perform outdoor activities rely on their boots to prevent them from slipping. However, choosing the best slip resistant footwear is challenging. Currently, there is no standard test method for evaluating slip resistance of footwear on ice surfaces. The SATRA STM 603 whole shoe tester is used in standard test methods (ASTM International, 2019b, F2913-19; International Organization for Standardization [ISO], 2019, 13287:2019) to evaluate the coefficient of friction (COF) of different shoes on other types of surface (e.g. wet and dry quarry tiles). This apparatus can be used in conjunction with a refrigerated ice tray to evaluate the COF of footwear on ice surfaces. The accompanying SATRA TM144:2011 is a proprietary mechanical method that provides rough guidelines for testing footwear on ice surfaces (frosted ice and smooth ice) with the SATRA STM 603. However, little information has been published about the validity of this method. Alternately, a human-centred method, called the Maximum Achievable Angle (MAA) test, was recently developed using the KITE Research Institute's WinterLab, located at the Toronto Rehabilitation Institute – University Health Network. This method evaluates footwear slip resistance on ice surfaces by measuring the maximum slope participants can walk up and down without slipping. This study was separated into three phases having the following objectives: phase 1A to refine the existing mechanical method by determining ice conditions using the SATRA ice tray, and phase 1B to evaluate the repeatability and reproducibility of the results obtained with this method using the SATRA STM 603 whole shoe tester on ice surfaces at two different laboratories: one at the Institut de recherche Robert-Sauvé en santé et en sécurité du travail (IRSST) and one at KITE; phase 2, to compare the mechanical method with the MAA method for evaluating footwear performance on ice surfaces; and phase 3, in the cases of inconsistencies between the two methods, to investigate which method is more reliable for ranking footwear by using another human-centred method. The two ice surfaces used in this study were based on the WinterLab’s ice surfaces. The WinterLab’s dry ice was a smooth cold ice formed and kept at -5.0 ± 1.0°C with ambient air temperature at 2.5 ± 2.0°C. The WinterLab’s wet ice was smooth melting ice formed and kept at -1.5 ± 1.0°C with ambient air temperature at 8.0 ± 2.0°C. For phase 1A, an ice preparation protocol was developed for the SATRA ice tray and a test protocol was defined based on existing standards. The monitoring of ice tray’s ice temperatures using thermistors revealed that the ice surface temperature fluctuated as a function of the refrigeration cycle of the ice tray. These fluctuations showed slightly different patterns between the IRSST and KITE labs. Thus, specific temperature set points and restricted temperature ranges for testing on dry and wet ice surfaces were determined for each lab to ensure that ice temperatures measured by the thermistors were as similar as possible in the two labs (within -6.0 to -5.0°C for dry ice, and within -2.0 to -0.5°C for wet ice) and closest to the KITE WinterLab’s ice temperatures. Any frost that formed naturally on the ice surface in ambient conditions was removed by wiping the ice surface with a wet cloth at the beginning of the tests. This helped ensure the SATRA ice surfaces better resemble the smooth surfaces of the WinterLab. For phase 1B, ten types of occupational footwear were tested at both labs, on dry and wet ice surfaces and in different slip modes. The results from the two labs for boots tested on wet ice were equivalent, both in terms of COF values and footwear ranking based on Bland-Altman analyses (Bland & Altman, 2010). For dry ice, although the footwear ranking was equivalent between the two labs, the COF values obtained at IRSST were systematically higher (by around 0.06) than those obtained at KITE. Limitations of this phase included an inability to control the temperature and the relative humidity in the two labs, which may have impeded the reproducibility of the mechanical method. In phase 2, each type of women’s footwear was tested by four female participants and each type of men’s footwear was tested by four male participants using the MAA method. The participants were asked to walk up and down slopes at their own pace on a 4 m walkway in the WinterLab, at KITE, while wearing a safety harness. The surface slope was increased systematically until the participants were no longer able to walk without slipping. Four MAA scores were recorded for each footwear model, each ice surface condition (dry or wet ice) and each direction (descending or ascending) defining the maximum slope the participant was able to walk up or down without slipping. The COF values and footwear rankings obtained on wet ice using the mechanical method were close to those obtained using the MAA method. However, for dry ice, the mechanical method gave a different footwear ranking compared with the MAA method. These observed differences may have been due to differences in ambient conditions that were out of our control (ambient air temperature and relative humidity). Efforts were made to maintain ice surface temperature for the mechanical tests as close as possible to the WinterLab’s ice surface temperature. The observed differences between the two methods may also be due to the mechanical method’s inability to simulate human gait. Hence, further research is needed to refine the mechanical method for estimating slip resistance performance, in order to improve agreement with human-centred approaches. In phase 3, two out of the ten types of occupational footwear were selected to be tested on wet ice in the WinterLab, with another human-centred method. A level walking test, developed by the KITE research team and using a passive motion tracking system to detect heel contact and toe-off from the velocity signal, measured the number of times each of five participants slipped while wearing a particular footwear model. On wet ice, the selected boots showed similar slip resistance when tested with the mechanical method (Phase 1B), while their slip-resistance qualities were significantly different when tested with the MAA method (Phase 2). The results of the level walking test, which consist in the number of slips encountered by the five participants during the test, were consistent with the results from the MAA method, and disagreed with the mechanical method. This study provided a better understanding of the use and limitations of the SATRA ice tray for measuring slip resistance. The results showed that our alternative mechanical method must be further refined to make its results more comparable to human-centred methods. Recommendations have been made to address this issue. This study also demonstrated that conducting tests on different ice surfaces, such as dry and wet surfaces, can be useful as a way of getting a more accurate picture of a boot performance, with both mechanical and human-centred methods.