Pakistan’s economy is based on agriculture that is highly dependent on water resources originating in the mountain sources of the Upper Indus. The aims of this study are to detect trends, variability analysis and assessment of changes in minimum (low), mean and maximum (high) flows in Mangla Basin, Pakistan. Trend analyses are performed by applying Mann-Kendall, non-parametric test. Sen’s method was applied to estimate slope time series that indicates changes in river flows. The influence of serial correlation was eliminated from time series by applying the trend-free pre-whitening (TFPW) method prior to the trend analysis. Results of this study revealed that trends were more common in mean and low flows compared to high flows. Statistically significant increasing and decreasing trends are noticed in different parts of Mangla watershed. The annual minimum flow at the outlet of Mangla watershed has decreased whereas mean and maximum flow has increased.