Dumb me, it was right under my nose in the next section.  In a nutshell, it looks like---if you know that both parents are carriers of the salt-wasting gene---the chances of having a virilised girl is 1 in 8.  With any other combination of parents, the odds would be somewhat less than that.  This is from the same article I posted above....

4. Other considerations for genetic counseling. The most common circumstance in which prenatal treatment is offered to a pregnant woman is when she and her partner have already had a child with CAH, in which case the likelihood of her bearing an affected girl with each successive pregnancy is 1/8.

Other scenarios may arise in genetic counseling. What if one partner has classic CAH and the carrier status of the partner is not known (recognizing that an affected mother will remain on glucocorticoid replacement regardless, but her dose may need to be increased to treat the fetus)? If the carrier frequency for classic CAH in the general population is 1.6% (equivalent to a disease frequency of 1/16,000, see Section IV.B), then the a priori likelihood of these parents giving birth to an affected girl is 0.4%, or 1/250 ([1 parent carrying 2 classic alleles] x [1.6% carrier frequency in general population] x [1/2 chance the carrier parent will pass his or her affected allele to the fetus] x [1/2 chance the fetus is female]).

Infants of mothers affected with the nonclassic disorder are also at slightly increased risk of developing classic CAH. Most studies examining genotypes in CAH (Section VI.I) did not specifically target nonclassic patients, and ascertainment biases and ethnic differences make it difficult to draw firm conclusions regarding allele frequencies in the nonclassic CAH patient population. Nevertheless, it appears that at least 50% of women clinically ascertained to have nonclassic CAH are compound heterozygotes for a classic and a nonclassic mutation (209, 210, 334). Accepting this figure, there is a priori approximately a 0.1% (1/1000) chance that a mother with nonclassic disease will give birth to a daughter affected with classic CAH ([50% carrier frequency of classic alleles among women with nonclassic disease] x [1.6% carrier frequency in general population] x [1/4 chance both classic alleles will be passed to the fetus] x [1/2 chance the fetus is female]).

In each of these scenarios, the risk of having an affected daughter is far less than the 1:8 risk with two known classic 21-hydroxylase deficiency carriers. Thus, prenatal therapy is not warranted unless the carrier status of the mate (as well as the genotype of a nonclassic patient) is first ascertained by hormonal testing and/or genotyping as part of preconception genetic counseling. This may be desired by some but not all couples. It may nevertheless be prudent to measure 17-OHP in such at-risk infants after birth, remembering that basal hormone measurements may not be sufficient to diagnose nonclassic CAH in infants and young children.