The objectives of this study were (1) to report the rates of new intramammary infection (IMI) and spontaneous IMI cure over the dry period in 3 dairy goat herds; (2) to evaluate the factors predicting infection dynamics over the dry period; and (3) to define milk quality parameter thresholds that predict infection dynamics over the dry period. Two consecutive udder-half milk samples were collected 10 to 14 d apart before dry-off from 288 goats in 3 herds, and 2 consecutive udder-half samples were collected 7 to 14 d apart in the following lactation, with the first sample being collected ≤10 d in milk, from 200 of the same goats. In 2 of the herds, udder-half milk samples were also collected at the same time points (n = 312 halves; 157 goats) for measurement of milk quality parameters. Standard aerobic culture of milk samples was performed for the detection of mastitis pathogens. To rule out the presence of Mycoplasma spp. IMI, milk samples were also cultured on modified Hayflick medium. Non-Mycoplasma isolates were speciated using MALDI-TOF mass spectrometry. Staphylococcal isolates, when not identified by MALDI-TOF, were speciated using partial gene sequence analysis of rpoB or tuf. When >1 sample from an udder half yielded the same species, available isolates from the first and last positive samples for that species were strain-typed using pulsed-field gel electrophoresis. Incidence of new IMI and cure rate were computed. Generalized linear mixed regression models were built to evaluate the associations between new IMI and pre-dry somatic cell score (SCS), between IMI persistence and half-level SCS, and between IMI persistence and pre-dry IMI species. Thresholds for pre-dry SCS and lactose concentration were computed to predict IMI persistence. Overall, 12.6% (48/380) of halves had a persistent IMI. Cumulative incidence of new IMI over the dry period was 13.2%, and cure rate was 52.0%. Pre-dry SCS was not associated with odds of new IMI or IMI persistence. Pre-dry IMI species was not associated with odds of persistence. Lactose concentration was not associated with odds of persistence. Regardless of culture data, the optimal pre-dry SCS threshold to detect IMI that would persist into the next lactation was 8.7, with sensitivity and specificity of 50 and 73.8%, respectively. Further studies on the effect of control measures on species-specific incidence and cure rates during the dry period are warranted.