A Bayesian procedure is presented to analyze test day performances in
the scope of Wood's lactation function: at(b) (erp (-ct)), where a, b,
and c = parameters, and t = time-dependent variable. The procedure ha
s been applied to 148,557 test day controls corresponding to records f
or 15,349 first lactations of Spanish dairy cows. The procedure uses a
ll available information and reduces the presence of atypical lactatio
ns, Moreover, the posterior marginal distribution of (co)variance comp
onents, breeding values, and systematic effects for the underlying var
iables of Wood's function are provided. The posterior means of the her
itabilities were 0.43, 0.17, 0.40, and 0.29 for the parameters a (init
ial milk yield), b (ascent to peak), c (descent from peak), and total
milk yield, respectively. Selection for total milk yield implies a gre
at effect on the initial milk yield (a) but only minor effects on asce
nt to peak (b) and descent from peak (c).