Introduction Factor XI (FXI) is a procoagulant element that circulates in plasma as a homodimer. It participates in both intrinsic and extrinsic coagulation pathways, as it is activated by both FXIIa and thrombin. Elevated levels of FXI are significantly associated with a high risk of thrombosis. FXI deficiency, considered a rare disorder in Caucasians, increases the risk of bleeding and protects against thrombosis. The identification of new naturally occurring genetic defects that cause FXI deficiency may help to identify new key domains and residues in the folding, secretion, dimerization and function of this molecule, as well as define its clinical impact. The aim of this study was to identify and characterize new genetic defects causing FXI deficiency. Methods We studied 91 unrelated Caucasian families with FXI deficiency (FXI:C<70%) identified mainly by prolonged aPTT. The entire F11 gene was sequenced by high-throughput sequencing (NGS). Structural variants (SVs), identified by MLPA, were characterized by nanopore long read sequencing using real-time enrichment methods on a MinION device, and validated by PCR and Sanger sequencing. Plasma FXI was characterized by functional methods using chromogenic and coagulation assays, and by immunological methods under both basal and silica-activated conditions. In addition, recombinant expression of wild-type (WT) and selected FXI variants was performed in HEK293. Results Thirty-one different F11 variants were identified. Three of them were novel SVs, being characterized at the nucleotide level by nanopore sequencing. They included two partial duplications, one involving exons 8-9 and the other exon 14; and one complete gene deletion spanning 7 MB including 142 genes. This was a de novo mutation originating in the paternal allele potentially during spermatogenesis. Interestingly, in all cases, repetitive elements were detected at the breakpoints. We also identified two small INDELs and 26 SNVs with different consequences (splicing defects, N=2; nonsense, N=3; and missense, N=21). We highlight 6 missense variants that, without affecting secretion (cross-reactive material, CRM+), produced FXI:C deficiency. Three of them were located in the catalytic domain; p.Pro538Leu, p.Glu565Lys and p.Cys599Tyr. Two variants disrupted the disulfide bridges p.Phe295Cys and p.Cys339Phe. The p.Phe295Cys was identified in a hemizygous state in a patient with severe FXI deficiency (FXI:C=4%). It produced two aberrant circulating FXI dimers through a potential shuffling of intramolecular disulfide bonds that were also observed in the recombinant protein. p.Cys339Phe affecting the residue involved in homodimerization was identified in 2 unrelated subjects with mild FXI deficiency. These individuals had FXI monomers in plasma that were activated by silica. Recombinant expression of this variant secreted only monomers, whereas WT expression in HEK293 secreted FXI dimers and monomers in the same proportion. The p.Arg268His variant located in the Apple 3 domain was identified in two subjects, one of them in homozygous state. Of note, we also detected 6 carriers of a variant affecting the same residue but causing CRM- deficiency (p.Arg268Cys). Most cases had not or only mild bleeding events which were not related to the mutation, type of deficiency, or FXI levels, although cases with severe deficiency presented more events. We point out a severe haemorrhage with fatal consequences during COVID-19 infection in a heterozyous carrier of the CRM+ p.Arg268His. Conclusions We confirm the genetic heterogeneity causing FXI deficiency in Caucasians. SVs affecting F11, probably caused by recombination of repetitive elements, are heterogeneous in type and extent and may not be detected by conventional short-read sequencing methods. Although most F11 variants cause type I defects, we have identified 6 type II or CRM+ deficiencies, and describe the first CRM+ variant affecting the Apple 3 domain. The data show that mutations in the catalytic domain or in aberrant disulfide bonds can allow secretion of inactive dimer variants. In addition, for the first time we describe the detection of monomeric variants in plasma. This, together with the similar secretion of WT FXI dimers and monomers in HEK293, suggests the relevance of Cys339 for the removal of WT monomers from the circulation.