The comparative study of the transformation among sediment phosphorus (P) fractions in different lake types is a global issue in lake ecosystems. However, interactions between sediment P fractions, environmental factors, and microorganisms vary with the nutrient status of lakes. In this study, we combine sequential extraction and metagenomics sequencing to assess the characteristics of P fractions and transformation in sediments from different lake types in the Inner Mongolian section of the Yellow River Basin. We then further explore the response of relevant microbial and environmental drivers to P fraction transformation and bioavailability in sediments. The sediments of all three lakes exhibited strong exogenous pollution input characteristics, and higher nutritional conditions led to enhanced sediment P fraction transformation ability. The transformation capacity of the sediment P fractions also differed among the different lake types at the same latitudes, which is affected by many factors such as lake environmental factors and microorganisms. Different drivers reflected the mutual control of weakly adsorbed phosphorus (WA-P), potential active phosphorus (PA-P), Fe/Al-bound phosphorus (NaOH-P), and Ca-bound phosphorus (HCl-P) with the bio-directly available phosphorus (Bio-P). The transformation of NaOH-P in reducing environments can improve P bioavailability, while HCl-P is not easily bioavailable in weakly alkaline environments. There were significant differences in the bacterial community diversity and composition between the different lake types at the same latitude (p < 0.05), and the role of P fractions was stronger in the sediments of lakes with rich biodiversity than in poor biodiversity. Lake eutrophication recovery was somewhat hindered by the microbial interactions of P cycling and P fractions within the sediment. This study provides data and theoretical support for exploring the commonalities and differences among different lake types in the Inner Mongolian section of the Yellow River Basin. Besides, it is representative and typical for promoting the optimization of ecological security patterns in ecologically fragile watersheds.