Abstract Serpentine-bearing sediment, a rare sediment type that is formed and deposited in divergent, convergent, transform, and collisional plate-tectonic settings, carries important evidence of sediment provenance. Specific sources of serpentine-rich sediment display grain assemblages of distinct character that can be used to infer the serpentinization condition and sediment formation. This study reports quantitative and qualitative results on serpentine components in sandstones from Ocean Drilling Program Legs 149 (Iberia), 210 (Newfoundland), and 125 (Mariana and Izu–Bonin regions), and from serpentine-rich debris flows and arenitic breccias in deep-water successions in the Northern Apennine fold–thrust belt. We propose a textural scheme that offers a visual guidance for evaluating serpentinite grains that can be broadly adopted, is easily reproducible, and reduces user bias in determining compositional modes that allow comparison of serpentinite grain populations in arenites from different depositional environments, provenance, and associated tectonic settings. These data allow us to define a scheme for serpentine-dominated deposits that demonstrates the presence of two main groups of grain textures (pseudomorphic and non-pesudomorphic) with specific mineralogy and crystal shape as a function of temperature and pressure in the source rocks. The quantitative analysis of the serpentine-rich arenites and fine-grained sediments derived from forearc and rifted continental-margin settings shows that the studied samples are characterized by high percentages (c. ≥ 80%) of serpentine detritus and subordinate dense minerals and other lithic fragments, including basalt. In rifted continental-margin settings, the prevalent textures in serpentinite sandstones consist of polygonal mesh, mesh-core, and hourglass that all belong to the pseudomorphic category, which preserves the pre-serpentine features and mineralogy. These textures are typically formed in low-temperature conditions (< 390°C); lizardite is the most common mineral, along with minor chrysotile and, in rare cases, antigorite. In contrast, in forearc settings, serpentine-rich grain assemblages exhibit dominantly non-pseudomorphic, interlocking, and interpenetrating textures, dominantly composed of lizardite and recrystallization of lizardite by antigorite. Minor preserved ultramafic minerals related to dynamic recrystallization might be associated with the diapiric rise and protrusion of serpentine bodies. The Northern Apennines case study adopted to test this model indicates that the relationship of detrital serpentine texture to setting can be employed in provenance studies. Firstly, serpentine-bearing sediments derived from ophiolites deformed in fold–thrust belts have more variable serpentinite content, ranging from a few percent to < 10% for samples from deep marine environments, to typically c. 20 to ≤ 50% for stream and beach samples. This compositional variation arises from mixing of sediments derived from deeper to shallower oceanic lithosphere (peridotites and serpentinites) with material from overlying volcanic rocks and sedimentary cover. The deep-water serpentine-rich sands of the Northern Apennines display variable compositions with intermediate characteristics. The source of the serpentine-bearing deposits is interpreted to be a residual oceanic lithosphere characterized by subcontinental mantle-lherzolite originated in the Middle–Late Jurassic by mantle delamination. The serpentinite-dominated debris flows and sand beds contain serpentine grains that exhibit compositional and textural transitions from pseudomorphic to non-pseudomorphic categories, along with changes in mineralogy from lizardite to antigorite. Serpentinite with pseudomorphic texture is observed in the mantle section away from the deformed area. On the contrary, the presence of serpentine-rich arenites with dominant non-pseudomorphic textures suggests derivation from tectonized serpentine along fault scarps and or as products of serpentine diapirism. The detailed serpentinite texture scheme used to classify sand grains in this study includes pseudomorphic (often lizardite, minor crysotile) and non-pseudomorphic textures, with the latter attributed to temperature- and pressure-controlled recrystallization (often to antigorite) or shearing during or after serpentinization. For comparison of different detrital-serpentinite populations, a new ternary plot is proposed where counted parameters are grouped into three end members: undeformed, deformed, and recrystallized. This plot appears to discriminate different sources of detrital serpentine by tectonic setting (e.g., Iberia and Newfoundland margins vs. Mariana forearc) and shows the potential complexity of serpentinite sources in the Apennine basin example. Additional texturally based petrographic data sets are needed to determine the usefulness of this plot in provenance studies.