With the aim of better understanding the tight gas reservoirs in the Zizhou area of east Ordos Basin, a total of 222 samples were collected from 50 wells for a series of experiments. In this study, three pore-throat combination types in sandstones were revealed and confirmed to play a controlling role in the distribution of throat size and the characteristics of gas-water relative permeability. The type-I sandstones are dominated by intercrystalline micropores connected by cluster throats, of which the distribution curves of throat size are narrow and have a strong single peak (peak ratio >30%). The pores in the type-II sandstones dominantly consist of secondary dissolution pores and intercrystalline micropores, and throats mainly occur as slice-shaped throats along cleavages between rigid grain margins and cluster throats in clay cement. The distribution curves of throat size for the type-II sandstones show a bimodal distribution with a substantial low-value region between the peaks (peak ratio <15%). Primary intergranular pores and secondary intergranular pores are mainly found in type-III samples, which are connected by various throats. The throat size distribution curves of type-III sandstones show a nearly normal distribution with low kurtosis (peak ratio <10%), and the micro-scale throat radii (>0.5 μm) constitute a large proportion. From type-I to type-III sandstones, the irreducible water saturation (Swo) decreased; furthermore, the slope of the curves of Krw/Krg in two-phase saturation zone decreased and the two-phase saturation zone increased, indicating that the gas relative flow ability increased. Variations of the permeability exist in sandstones with different pore-throat combination types, which indicate the type-III sandstones are better reservoirs, followed by type-II sandstones and type-I sandstones. As an important factor affecting the reservoir quality, the pore-throat combination type in sandstones is the cumulative expression of lithology and diagenetic modifications with strong heterogeneity.