Spatiotemporal patterns of soil moisture are important for understanding landscape hydrologic processes. While the spatiotemporal characteristics of soil moisture content (θ) have been frequently studied, those of soil matric potential (ψ) remain sparse. We investigated the spatial variability and temporal stability of ψ at multiple depths (10–100 cm) across a 7.9-ha forested catchment and their relationships with soil type, terrain, and season. The results from a 5.5-yr database consisting of 62 sites clearly showed a downward parabolic trend in ψ spatial variability with decreasing spatial mean ψ values (i.e., becoming drier) across all depths. The catchment’s overall spatial variability of ψ generally increased with soil depth and was relatively high during summer and fall. Sand and silt contents were significant factors (p < 0.05) influencing ψ in surface layers (0–20 cm), while ψ values in the subsurface (40–100 cm) were highly correlated with elevation. The temporal stability of ψ spatial pattern was generally higher in the surface soil than the subsoil but weaker in the spring than in other seasons. Moreover, relatively dry areas tended to have less evident ψ temporal stability at each depth, attributable to a higher sensitivity of ψ to changes in θ when the soil is dry. Finally, at least one of the representative sites of the catchment mean ψ at each depth was found within a south-facing concave hillslope at mid-elevation, and these sites were different from the representative sites of the catchment mean θ. The results of this study have implications for upscaling soil water from point-based observations to hillslope and catchment scales.