The introduction of 'solids' (i.e. complementary foods) to the milk-only diet of early infancy affects the development of the gut microbiota. The aim of this study was to determine whether a 'baby-led' approach to complementary feeding, that encourages early introduction of an adult type diet, results in alterations to gut microbiota composition compared to traditional spoon-feeding. The Baby-Led Introduction to SolidS (BLISS) study randomized 206 infants to BLISS (a modified version of Baby-Led Weaning (BLW): introduction of solids at 6 months of age, followed by self-feeding of family foods) or Control (traditional spoon-feeding of purées) groups. Fecal microbiotas and 3-day weighed diet records were analysed from a subset of 74 infants at 7 and 12 months of age. The composition of the microbiota was determined by sequencing 16S rRNA genes amplified by PCR from bulk DNA extracted from feces. Diet records were used to estimate food and dietary fiber intake. Alpha diversity (number of Operational Taxonomic Units; OTUs) was significantly lower in BLISS infants at 12 months (difference, 95% CI: 31, 3.4 to 58.5; p = 0.028) and, while there were no significant differences between Control and BLISS infants in relative abundances of or at 7 or 12 months of age, OTUs representing the genus were less prevalent in BLISS microbiotas at 12 months. Mediation models demonstrated that 'fruit and vegetables' and 'dietary fiber' intake explained 29% and 25% respectively of the relationship between group (BLISS versus Control) and alpha diversity. The introduction of solid foods (complementary feeding, weaning) to infants leads to more complex compositions of microbial communities (microbiota, microbiome) in the gut. In Baby Led Weaning (BLW), infants are given only finger foods they can pick up and feed themselves - there is no parental spoon-feeding of puréed baby foods - and infants are encouraged to eat family meals. BLW is a new approach to infant feeding that is increasing in popularity in the United States, New Zealand, the United Kingdom and Canada. We used mediation modeling, commonly used in health research but not until now in microbiota studies, to identify particular dietary components that affected the development of the infant gut microbiota.