The diatom genus Pseudo-nitzschia includes many toxigenic species that produce domoic acid, the causative agent of amnesic shellfish poisoning, making accurate species-level detection essential for harmful algal bloom monitoring. ITS1 is widely used in Pseudo-nitzschia metabarcoding and automated ribosomal intergenic spacer analysis (ARISA), but its intragenomic variation and structural evolution remain poorly resolved, complicating interpretation of environmental DNA datasets. Here, we combined high-throughput ITS1 amplicon sequencing, isolate-level variant profiling, thermodynamic RNA folding, and homology modeling to evaluate ITS1 diversity in Pseudo-nitzschia. We analyzed 150 isolates representing 18 species from Chinese coastal waters. Nearly all isolates contained one dominant amplicon sequence variant (dASV) with multiple non-dominant variants (ndASVs), demonstrating pervasive ITS1 intragenomic variation. ASV numbers varied among isolates and species, and several species showed multiple dASVs among isolates, indicating that ITS1 captures intraspecific genetic diversity. However, ndASVs often matched dASVs from other isolates of the same species, other Pseudo-nitzschia species, or even other genera, revealing a risk of overestimating diversity or misassigning taxa if non-dominant variants are not carefully considered. Simulated ARISA profiles showed that ITS1 length polymorphism reflects species- or clade-level patterns, supporting ITS1-based ARISA but requiring caution for closely related taxa with overlapping profiles. To clarify the biological meaning of ITS1 variation, we established a universal consensus ITS1 secondary structure model for Pseudo-nitzschia. The model revealed a conserved architectural core with major helices and signature motifs, together with variable distal domains such as Helix IIa. Comparative structural mapping showed hierarchical divergence across evolutionary clades. Conserved domains provide a framework for filtering putative pseudogenes from environmental datasets, whereas localized structural polymorphisms and compensatory base changes can serve as molecular fingerprints for geographically differentiated populations. In P. delicatissima, P. galaxiae, P. multistriata, and P. simulans, spatially structured ITS1 variants suggest population divergence and potential early-stage cryptic speciation. Overall, our results support ITS1 as a high-resolution marker for Pseudo-nitzschia detection, while showing that reliable metabarcoding interpretation requires integrated consideration of dominant variants, intragenomic variation, length polymorphism, and RNA secondary structure. This sequence-structure framework improves molecular taxonomy and eDNA-based monitoring of toxigenic Pseudo-nitzschia. To further improve the resolution of Pseudo-nitzschia species and track the evolutionary trajectories of ITS in Pseudo-nitzschia species, we are also conducting full-length ITS analysis.