Hello to all who have helped me on this topic, first I need to apologize for apparently replying only now... In fact I use the "Pan" Newsreader to read the list, and I posted a reply to the thread a week after your suggestions through Pan, and I only now realised that the posting never arrived on the list although Pan gave me no error message at all! So, let me try again using the good old email-to-email way. You all helped me so much! This is what I'm doing now: - I separated my large file into several chapter files - In each file, I include a Sweave options file using "\SweaveInput" - I make sure to first run a "pre" file, which checks if some files containing data from the database that I need repeatedly are there or not, and if they are not too old. If the files are absent or expired, the database is queried and the files recreated. - In each file, I then "source" an init.R file, which reads the previously created files and sets some "global" variables I'll need all the time. So, the querying is done maximum 1 time, while the file reading and variable setting is re-run for each chapter, which doesn't seem to be a problem (performance-wise). - In my master tex file, I include the different chapters. If I want to generate PDF for only a single chapter, I cannot use the "\includeonly" directive, because I will always need to run "pre" first, and then the chapter I want. So, I just comment out the things I don't want to run. - In "my" Makefile (it's Mark's really, with some minor adaptations), I specify the following to make sure "pre" is run first before running the individual chapters: RNWFILES = pre.Rnw intro.Rnw $(wildcard c*.Rnw) So, with all this I can get "whole document" PDFs or "per-chapter" PDFs - Great! Actually, then I go on and feed the tex file(s) to "latex2html", a great tool, to generate HTML equivalents all ready with navigation, buttons and all. Maybe it's not yet perfect, but for sure it is much, much better than before! Greetings from Luxembourg, Anne-Marie