In recent years, a new trend of seasonal labour migration from Armenia to Russia has emerged. Based on a novel household survey, this paper analyses how successful seasonal migrants are in increasing their incomes. Applying matching operators allows addressing endogenous self-selection to migration. We identify negative selection based on education, employment and pre-migration income. This is reflected by a premium for low skills in Russia relative to Armenia, luring seasonal migrants into low-skill jobs, mainly in the construction sector. The income gain for a migrant is estimated at $ 480 relative to the approximately $ 50 that the same individual would have earned in Armenia. The results are robust to various matching techniques and specifications.