The cat flea, Ctenocephalides felis felis, is the major initiator of flea bite hypersensitivity in dogs. Previous analyses of whole extracts of the flea and flea salivary secretions have failed to identify the allergens responsible. We dissected >2000 salivary glands from adult female fleas, extracted them into buffered saline containing protease inhibitors and fractionated the extract using gel permeation HPLC. Dogs were classified as hypersensitive to fleas (flea-feeding positive, FF+) or insensitive (flea-feeding negative, FF−) using a provocative test with live fleas. The allergenicity of the components of the salivary gland extract was tested by intradermal injection of samples of the column eluates. Dogs were also injected intradermally with a sample of whole salivary gland extract, and with histamine as a positive control. Negative control injections consisted of eluate from the column collected prior to fractions containing any protein. The skin of FF− dogs either did not respond or had a minimal response (a bleb ∼2 mm larger than the injection blebs at the negative control injection sites) to all fractions and to the whole extract; histamine control injections produced positive responses (defined as wheals 5 mm greater than the blebs at the negative control injection sites) in all dogs. The skin of three of the nine FF+ dogs reacted positively to injection of a fraction containing protein/s with apparent MW 40k. Five other FF+ dogs reacted positively to the fractions containing proteins with apparent MW 12-8k. A single dog responded with very large, red wheals to injection of both the ∼MW 40k and MW12-8k fractions. These findings suggest that proteins with apparent MW 40k and MW 12k-8k are important in flea bite hypersensitivity. This work also supports a previous finding that mice which had been exposed to flea bites had antibodies to proteins with ∼MW 40k that were detected in salivary secretions of the flea.