Summary For northern and arctic regions, the spring break-up period has important socio-economic, ecological and morphological effects. While these impacts are reasonably well understood, spatial and temporal assessments of break-up timing and duration remain limited due to the lack of readily available hydrometric data. For this study, the Mackenzie River Basin (MRB) of Canada is selected as a test watershed in which the spatial and temporal aspects of observed (1913–2002) spring river-ice break-up are characterized. Data from 29 Water Survey of Canada gauging sites are used including the commonly assessed ‘Last B date’ (last ice effect) and two hydrometric variables extracted directly from original water-level recording charts (the timing of initiation of break-up and peak water-level during break-up). It is found that the extracted variables provide a more physically based quantitative description of the break-up season in the MRB compared to the ‘Last B date’ method. On average, the northwards progressing ice break-up season within the MRB lasts ∼8 weeks but historically has varied within a window representative of ∼3 months of the year. The break-up period at specific locations varies from 4 days to 4 weeks. Results also indicate an anomalous zone of earlier spring break-up in the upper Peace and Athabasca region that may be partially related to the effects of flow regulation. In addition, the Mann–Kendall test reveals significantly earlier trends in the timing of spring break-up (∼1 day/decade) in upstream portions of the major tributaries of the MRB over the period 1970–2002. While similar trends have been found for other hydroclimatic variables in the basin, this study highlights the temporal patterns and variability of the spring break-up period in the Mackenzie River system.