Sounds like operator error. If you use a MathaCAD variable, it has units. So if you define "Len:10", "Width:15" and then in your equation you mistakenly say "A:L*W" it will think that your "A" is redefining Angstrom to volume ("L" is "liter") times power ("W" is "Watt". I used to do that all the time. I try to stop myself from doing it by never using a simple variable that could be a unit. For example, my "Area" terms are always "A.something" since none of the built-in functions use the "dot" subscript operator.
So if you say "L.yard:10", and "W.yard:15", then "A.yard:L.yard*W.yard=150" with no units.
Of course, not using units is cutting out a huge proportion of the power of MathCAD. For empirical equations where the units won't work out I strip the units inside the equation. For example (made up equation that doesn't mean anything):
MU.gas:[π]/4*(ID.pipe/in)^2.5*ln((ID.pipe/in)/L.pipe/mi)
This lets me input pipe ID in mm (or whatever length unit I choose that day), pipe length in ft, and MathCAD just uses the right magnitude in the empirical equation. I have some flow equations that I looked up 15 years ago, probably can't even find the reference today and before I started using this technique I had to remember the units of all the inputs. If someone asked me today to tell them the friction drop of 3 km of 200 mm pipe with 800,000 m^3 of gas flowing at 18 bar at the head, I can just input the given data without worrying about the units that my empirical equation needs--saves a ton of work. I've rewritten all of these old equations in this unit-neutral method and get far more consistent results.
David