I shifted to lead-free production in late 2004, with the last leaded production run taking place in early 2006.
I started researching realiability of lead-free soldering with the very first version of the Catweasel MK4 (the red one, first prototype in october 2004). After fiddling with temperature curves and stencil thickness for quite a while, I figured out parameters that have even higher reliability than the leaded solder process I used before (I was using too thick stencils before).
The biggest difference is in humidity requirements of the chips: With the higher peak soldering temperature, you really should have under 5% humidity in all chips, so either they are unpacked on-site and assembled&soldered within 6-12 hours, or you bake them until "right before assembly". Humidity in the chips while soldering results in microcracks in the plastic packages, which leads to long-term failure after 6-12 months. The "bake before assembly" step is one of the biggest changes that I have introduced just before the mass-production of Catweasel MK4. The Catweasel MK4 was the first with under 8 per-mille return rate.
I do have my own recepie of mixing new solderpaste with solderpaste that has been opened a few days before (not longer than two weeks) in order to reach better consistency, but that's something I'd only want to share with someone who is paying a good consultant fee. I gathered this knowledge in years of trial-and-error, made hundreds of macro photos of solder whiskers during the optical check phase after soldering and adapted the whole production process with the gathered knowledge. Please understand that I don't let go of this knowledge for free.