Brief introduction to bulk hard disk encryption: The data is encrypted sector-per-sector. The underlying cipher operates in CBC mode with the IV set to the relative sector number. If an attacker finds two sector starting with the same cipher text, he is able to deduce the difference between the plain texts by examining the difference of the IV. Since the IV is known, this can be done. If one of the plain text blocks is zero, the difference yields the original content. In no case this attack can be used to derive the encryption key.
Because only the first block of a sector is examined, the chance of finding an identical block is 2^128 (with a 128 bit block cipher). Because the number of possible comparisons develops as an arithmetic series in n, the number of sectors, the chance of not finding two identical blocks is given by (1-p)^(n(n-1)/2) with p=2^128. With n of the magnitude of 10^9 this chance is almost 0. The lower bound can be approximated by the Bernoulli inequality (1+x)^a = 1 + ax. Assuming a to be 10^18 this results in 1 - 2.93e-21. So in the worst case, the probability, that there are no identical cipher blocks, has the magnitude of 1 - 2.93e-21. (Thanks to Pascal Brisset for pointing out this mistake)
For a modern 200 GB disk, the difference between the IVs can be ~2^29. In the worst case (the second block = zero) the first 29 bits of a 512 bit block are revealed.