Data Transformation Services been completely re-written for SQL Server 2005. The main difference is the separation of workflow and dataflow. But additional niceties include for/foreach looping, debugging, and OR constraints which mean daft hacks are not required. In fact modifying the package at runtime is impossible.

The new dataflow pipeline deserves a special mention as its no longer one row in one row out. The transform can consume data from multiple sources and send to multiple destinations in a single task - this allows for some interesting merging and splitting of data. Set operations are also supported - sort, aggregate,  split, multicast amongst others.

DTS has grown up and can play with the big boys. If DTS 2000 is not enough for your needs right now then consider waiting for SQL Server 2005 and not trying to implement yourself.