Errata for the second printing, first edition
(all of these are also in the first printing)

Page 10, table 2.2: The hazard at times 11 and 12 should be 1/26 and
1/9.  The other columns are all correct.  The 1/8 also appears further
down the page.

Page 23, line 11: the final expression should be (t-10) rather than (t+10)

Page 28, equation 2.15: remove the final "ds".

Page 33, the unnumbered equation on line -5: The leading \Lamda should have
"D" as a subscript.

Page 50, equation 3.13: the denominators for the equation should be 1-\lambda,
not 1+\lambda.

Page 73, line 1: Although the Splus code says "method='breslow'", the output
that is shown is for the Efron approximation. 

Page 121, line 8-9: G is non-negative definite, not positive definite.

Page 135, fit.vet printout: The veterans data posted on my web site,
and used in this particular output, was incorrect.  Observation 1 of the
data had inadvertently been removed from the data file (when trimming
off some introductory material).  The remaining text in the book, including 
the cox.zph printout, is correct.  

Page 203, SAS output: The labels for the size1-size4 and number1-number4
variables are reversed.

Page 219, table 8.15: The last two values in the table are transposed.
There are 0 subjects with 6-9 entries into the postsurgery remission state,
and 1 with 10+ entries.

Page 219, line -4: This states that all 42 of the possible transitions
occur.  Actually, only 40 of the 42 are observed, and 4 others have only
a single instance.

Page 229, table 8.16: Due to an error in my sas program to read the colon
data, all of the recurrence --> death times were missing.  The program on
the web page has been repaired.  The correct coefficients for the alternate
model are -.022, -.382, .475, and .804, respectively.  None of the conclusions
from the fit are changed.

Page 252, line 4.  Inconsistent variables (s and t) in the integral.

Page 283 and 285, figures 10.8 and 10.9.  The legends are incorrect for
UDCA and Placebo; placebo is the solid line in both cases.

Page 321, line 3: '90 104' instead of '0,104'