ABSTRACT Wet (dry) spells can cause extreme climatic conditions, such as floods (droughts) which can adversely influence the natural resources. A spatio‐temporal analysis of observed wet and dry spells is carried out for vulnerable and data sparse region of Pakistan (24°–38°N and 61°–78°E). Observed monthly precipitation data sets from 46 weather station are used for a length of consecutive 32 years (1976−2007). Additionally, after bias adjustment, the Asian Precipitation Highly‐Resolved Observational Data Integration Towards Evaluation (APHRODITE) precipitation data set, abbreviated as APH, is utilized to corroborate the findings. For both data sets, a threshold of 1 mm is used to define a wet spell. Decadal variability of precipitation for observed and APH data sets indicates that there is gradual decrease in wet spell length for arid and humid regimes, as compared to semi‐arid regimes where there is no change in wet spell length. Monthly dry and wet slope difference (SD), on a log–log plot between number of spells and spell length, is used to classify precipitation regimes in Pakistan, for the first time. A weather station is categorized as humid if SD is less than −2.38 in units of number of spells per length of spells. If SD lies between −2.37 and −0.51, then the weather station is classified as semi‐arid and if SD is greater than −0.51, then it is classified as arid. Thus, according to SD classification, 66% of area in Pakistan is arid, whereas 30% (4%) is semi‐arid (humid). Comparison of precipitation regimes based upon observed and APH data sets with other climate classification schemes that involve both precipitation and temperature is presented. Monthly dry and wet slope difference (SD), extracted from log–log plot between number of spells and spell length, is used to classify precipitation regimes in Pakistan, and is displayed in upper row panels using observed and Asian Precipitation Highly‐Resolved Observational Data Integration Towards Evaluation precipitation data sets. The Erniç aridity index and Köppen classification technique‐based results are compared in lower row panels. Weather station locations are displayed using open circles.