International migration is one of the major determinants of demographic change. Although efforts to produce comparable statistics are underway estimates of demographic flows are inexistent outdated or largely inconsistent for most countries. We estimate age and gender-specific migration rates using data extracted from a large sample of Yahoo! e-mail messages. Self-reported age and gender of anonymized e-mail users were linked to the geographic locations (mapped from IP addresses) from where users sent e-mail messages over time (2009-2011). The users' country of residence over time was inferred as the one from where most e-mail messages were sent. Our estimates of age profiles of migration are qualitatively consistent with existing administrative data sources. Selection bias generates uncertainty for estimates at one point in time especially for developing countries. However our approach allows us to compare in a reliable way migration trends of females and males. We document the recent increase in human mobility and we observe that female mobility has been increasing at a faster pace. Our findings suggest that e-mail data may complement existing migration data resolve inconsistencies arising from different definitions of migration and provide new and rich information on mobility patterns and social networks of migrants. The use of digital records for demographic research has the potential to become particularly important for developing countries where the diffusion of Internet will be faster than the development of mature demographic registration systems.