Mechanism Assembly

My first thought is either improper dependencies/references or model construction.

The first can be confirmed by exporting to neutral files and re-importing, constructing the mechanism and checking if the model then works.

The second is difficult to communicate and is better with some idea of the way your model was constructed, how the joints were made in what components. 

One tip which you likely know is keep all parts in a rigid body in their own assembly - or never rigidly attach a part to another part with a joint to create a multi-part body.  Also don't define joint reference from different parts in the same rigid body assembly.  Instead create assembly level datums to make the connections with or have the references all in one part of the rigid assembly. (avoid multi-part references)  For example don't define a pin joint where the axis is in one part and the Z reference is in another.  For another example if a body has two pin joints but on two different parts of the body, place the pin joint references (axis, datum planes) at the assembly level, don't even define the two pin joints from different parts within the rigid body assembly.

Let me know if some of what I am saying needs further explanation.  Anything more about the model you can give would help, screenshots, model tree snips, error messages, how you are checking/confirming bodies and masses, etc.

First off, if you wish to run a "Dynamic" analysis, ALL parts MUST have a density assigned to them so they each have a mass - there are no exceptions to this one. Second, make sure that every part in your all sub-assemblies are fully constrained within their own assembly - never leave any "Allow Assumptions" for any individual part. Also, make sure that every part that gets connected is so formally" - this means while you build each joint, only use the same 2 parts while selecting references - even if one part in your "ground" body is rigidly connected to another "ground" body. I have seen issues with this in the past.

Good Luck!