This observational study addressed a critical gap in the understanding of the precursors of infant attachment by examining whether a new conceptualization of maternal caregiving behavior, secure base provision (SBP), explained variance in attachment above and beyond variance explained by sensitivity. Participants included 83 low‐socioeconomic status (SES), 4.5‐month‐old infants (56% male) and their mothers. Infant–mother dyads completed laboratory tasks at 4.5 months and three 30‐min home visits between 7 and 9 months, then returned to the laboratory at 12 months for an attachment assessment. Maternal sensitivity did not significantly predict infant attachment security. SBP significantly predicted infant attachment, over and above sensitivity, with an effect size eight times larger than that of sensitivity in meta‐analytic findings for low‐SES families.