The objectives of this thesis were i) to compare lifetime sow performance and number of piglets born alive (NBA) across parities, according to 3 NBA classifications in first and second parity and ii) to determine the number of parities a sow should be retained in a breeding to wean system to maximize returns over total costs per weaned pig and net return on investment. Sensitivity of returns over total costs per weaned pig to feed price and number born alive (NBA) was also analyzed. For both objectives the production data used were collected between the years 2001 to 2014 at 17 farms owned by the same Mid-West U.S. company. A total of 502,491 records accounting for the lifetime performance of 105,719 sows were used in this analysis. Data included both purebred and crossbred sow information. For the first objective sows were classified into 3 NBA categories (i.e., low, medium, and high) according to the 25th percentiles of NBA in parity 1 and parity 2. Mixed model analyses were applied to the data. Sows in low first and second parity NBA classification had an average of 1 to 1.8 NBA less per parity in parities 3 through 7 when compared with sows in the medium and high NBA classification, respectively (P < 0.05). Conversely, sows classified as high NBA in parity 1 and parity 2 had greater NBA in all subsequent parities as well as total lifetime NBA when compared with sows classified as low or medium NBA (P < 0.05). The results indicated that with the inclusion of second parity classification, sows classified as low NBA in parity 1 had a greater level of variation between later parity estimates than sows that were classified as high NBA in parity 1 (P < 0.05). Thus, the effect that parity 2 classification had on estimated later parity performance was found to be dependent on parity 1 classification (P-interaction < 0.05). The interaction between classes was seen when predicting total lifetime performance on a number born alive basis as well. As parity 1 classification increased, the difference between estimates by high and low NBA parity 2 classification became smaller. It was shown that parity 1 and 2 classification only had a small effect on the parity of removal. In conclusion, it was demonstrated that the use of first and second parity performance (based on number born alive) can be used effectively to predict subsequent parity and lifetime performance which can be used to aid in selection and culling decision early in the sows’ life. For the second objective, projected budgets were used to compare returns for various parity distributions. A “steady-state” model was used to demonstrate returns based on an existing farm versus a system that is just entering production. The returns over total cost per weaned pig were calculated using both variable and fixed costs associated with a 5,000 head sow farm based on the proportion of sows by parity in the distribution. Estimates for NBA and pre-wean mortality were shown to be statistically different between parities (P < 0.05). The parity distribution where sows are culled after parity 6 produced weaned pigs at lowest costs with the highest returns. Keeping sows until later parities, such as parity 7 through 9, had greater returns over total costs versus culling sows after parities 1 through 4. Net return on investment was maximized in the parity distribution that culled after parity 6, and parity distributions culling after parities 5 through 8 all showed approximately a 15% return on investment. Sensitivity analysis results indicate that as sows increased in NBA, culling after parities 5 through 9 saw higher returns. When feed prices were low the parity distributions culling after parities 6 through 9 resulted in the greatest returns and when feed prices were high, all parity distribution’s costs exceed returns per weaned pig, but culling after parities 5 and 6 were the parity distributions that were closest to break-even. Results from this body of work demonstrate the economic benefits associated with increased sow longevity as well as a means to select superior dams early in their life. Both production strategies have the potential to be implemented into breeding herds to increase the profit realized by the system.