Recent progress in sequencing technologies has made it possible to investigate the role of rare variants in disease etiology. Tests based on affected sib pairs can be more powerful than for case-control designs, as rare variants tend to be enriched in families. The main idea for such a test is that rare variants will be found preferentially on haplotypes shared identical by descent (IBD) versus haplotypes not shared IBD. We construct a couple of tests statistics which measure the departure of the joint distribution of allele count (0-4) and IBD state (0,1, or 2) under ascertainment (two sibs affected) from what it would be under simple Mendelian transmission. We show that these tests have good type I error properties, and at least one of them is more powerful than existing relevant methods for testing association. Finally we extend our methods to test all rare variants in a pathway, as opposed to in a single gene or region, an attractive feature not available with current methods.