As part of the national Alzheimer’s Disease Sequencing Project (ADSP), genome-wide linkage analysis was performed in 111 multi-ethnic families multiply affected by Late Onset Alzheimer’s Disease (LOAD), with admixture mapping also carried out on Hispanic pedigrees. WGS data were generated for 578 subjects (affected and unaffected members) from 111 families (67 Caribbean-Hispanic [CH], 44 non-Hispanic White [NHW]) and 30 multigenerational Dutch families (DF). Based on linkage support from multiple CH families, genomic regions on 7p14, 11q12, and 3q29 (based on one CH family), were prioritized for WGS analysis; regions on 1q23, 14q32, and regions on chr1, 4-6, 10, 16 and 18 (because of strong results in single families) were prioritized in the set of NHW families. In DF, AD is linked to chromosome 5p14.3. Results from admixture mapping prioritized a region on chr13 in the CH families. Prioritization of the sequence variants was performed using multiple criteria including: segregation with disease, allele frequency, functional annotation, and other datasets such as the ADSP case-control dataset. Four missense variants under 7q14 and 5 variants under 11q12 segregate in at least 1 CH family, including GHRHR and TMEM132A. We identified several intronic variants in the 3q29 region segregating with LOAD in multiple CH families. Further annotation of the non-coding variants suggests regulatory features. Analysis of the NHW families has identified 19 genic and 11 intergenic variants of high priority. Three of these are NOS1AP gene variants; the variants segregated with disease (including one with CADD score of 13.6), were absent from unaffecteds, and NOS1AP showed nominal disease association in the ADSP case-control dataset (p=0.049). Additional intergenic variants segregated with disease and likely influence transcription factor binding (CATO annotation). Within the 15q14.3 region, 2 deleterious variants (PRDM9 and Drosha) and 5 intergenic variants with a CADD score>15 were found to be shared in by all relatives of the major DF. Additional analysis is ongoing. These results suggest that both coding and non-coding rare variants influence LOAD susceptibility, and show the continued utility of family-based approaches. Variants identified are currently being validated using orthogonal technologies, and follow-up analyses are being conducted.