Daily feed intake during lactation was recorded on purebred Yorkshire (n = 1587), Landrace (n = 2197), and F1 Yorkshire x Landrace (n = 6932) litters from day 1 to 22 of lactation. Lactation feed intake (LFI) curves were predicted using a mixed model which included fixed effects of breed, season, parity group (PG), day of lactation, interactions of day with breed and PG, and a covariate for litter size after cross–fostering. Random effects included litter, contemporary group (herd–year–month), dam, and sire nested within breed. Least squares means for each day were used to define LFI curves by breed through day 22 of lactation. Yorkshire and Landrace LFI curves were not different (P = 0.09), though both differed from the LFI curve (P < 0.05) of F1 sows. Evaluation of the difference in feed intake between 2 consecutive days (DC) of lactation resulted in the following classifications: 3 periods for purebreds, day 1 to 6 (PB1), day 7 to 10 (PB2), and day 11 to 18 (PB3); 2 periods for F1 sows, day 1 to 5 (C1) and day 6 to 18 (C2). Average rate of change in intake (ARC), average daily intake (ADI), and variation from predicted LFI values (VAR) metrics were estimated for each period in both purebred and F1 sows. The lowest ARC and ADI metrics, but highest VAR (P < 0.05) in each period of lactation occurred in PG 1 sows and during the summer. An increase in ARC and ADI metrics resulted in higher 21–day litter weaning weights (P < 0.05), while decreasing VAR metrics late in lactation (PB3 and C2) resulted in higher 21–day litter weaning weights (LW21) and shorter wean–to–first–service intervals (P < 0.05). Heritability estimates for average daily lactation feed intake (ADFI), average rate of change (ARCT), and variation from predicted daily LFI values (VART) calculated over the entire lactation were 0.37, 0.24, and 0.16, respectively. Heritability estimates were variable across periods of lactation for ARC (0.03 - 0.17), ADI (0.09 - 0.36), and VAR (0.04 - 0.18) in purebred and F1 populations. Average rate of change, ADI, and VAR metrics during each period of lactation were most highly correlated with the corresponding metric over the duration of lactation. Genetic covariance estimates for each LFI metric with adjusted backfat depth (ADJBF), adjusted loin muscle area (ADJLMA), and days to 113. 4 kg (DAYS) were obtained with bivariate animal models. Genetic correlations between LFI metrics in first parity sows and performance ranged from –0.22 to 0.37 with ADJBF, –0.43 to 0.06 with ADJLMA, and –0.68 to 0.16 with DAYS. Deviations from predicted LFI values were quantified using an internally studentized residual (SR). A SR ≤ –1.71, equivalent to observed LFI at least 1.9 kg less than predicted, was considered a significant negative deviation from predicted LFI values (DEV). Zero DEV occurred in 60% of lactation records, and 34% of negative deviations occurred during the summer months (June, July, August) which was more frequent when compared to the spring (26%), fall (23%), and winter (17%) months. Adjusted 21–day litter weaning weight decreased as the number of DEV increased within a single lactation period, and wean–to–first–service interval (WTSI) increased when at least 3 DEV occurred. An increase in DEV during early lactation did not affect LW21 or WTSI (P > 0.05), though an increase in number of DEV after day 5 of lactation was associated with lower LW21 and longer WTSI. If a DEV occurred the prior day, a DEV was 8.7 and 39.5 times more likely to occur than if a DEV had not occurred the previous day for purebred and F1 sows, respectively. In purebred sows, subsequent number born alive (NBA) was smaller when ≥ 5 DEV had occurred during the previous lactation, though NBA was relatively unaffected (P > 0.05) by DEV in F1 sows. The number of DEV during the previous lactation did not affect (P > 0.05) subsequent number stillborn (NSB) in purebred sows, but NSB significantly increased when ≥ 3 DEV occurred in F1 sows. Managing sows during lactation continues to be an essential component of productivity, and feeding sows with the intention of meeting predicted lactation feed intake levels is a valuable tool which should reduce the occurrence of inadequate intake and maximize reproductive performance. There is potential to alter the shape of the lactation feed intake curve through direct selection for average rate of change in intake, average daily feed intake, and variation from predicted lactation feed intake values during any period of lactation. Including lactation feed intake metrics in a maternal selection index should permit simultaneous improvement in sow productivity, grow–finish performance, and lactation feed intake.