Sue Dynarski claims to teach econometrics, but in practice they just read journal articles and do pointless (and lengthy) Stata problem sets. Students leave knowing nothing of metrics and love her for it. This is in her graduate level courses.
Teach econometrics to grad students as a reg monkey

What? Most people can understand geometric arguments which is so abstract?
Honestly use Ruud's book. It gives a nice geometric argument that most people can understand and will make fielding questions much easier. Hayashi will just bog you down with his love affair with the GMM framework and the lack of appendices overviewing the prerequisites.
Pictures dude. Pictures

Are you talking about Notredame? Bill Evans teaches this way. It's a disgrace.
I know of a decent uni where a ref monkey teaches metrics. The course is heavy on Mostly Harmless type stuff and thinking through endogeneity problems.
I am a reg monkey, and IMO first year metrics should be theory based. No, I cannot prove that convergence a.s. implies convergence in probability, but I think surviving the classes helped me a lot with better understanding what my reg y x, r is doing. My concern has long been that people who are only taught Mostly Harmless type stuff have no idea what they're doing in Stata.

Thanks everyone. The guy who taught the course before me used Gujarati. Would you switch to Wooldridge?
He used Gujarati at Grad level? I thought he only had a UG book.
Wooldridge is very good but, as it says on the tin, only has cross sectional and panel metrics. 
what about Wooldridge's introductory econometrics book? Or Davidson and MacKinnon? Btw this guy used Gujarati's UG book....I teach in a UVLRM to people with very limited math background, I guess this course will end up being similar to an advanced undergrad course
He used Gujarati at Grad level? I thought he only had a UG book.
Wooldridge is very good but, as it says on the tin, only has cross sectional and panel metrics. 
what about Wooldridge's introductory econometrics book? Or Davidson and MacKinnon? Btw this guy used Gujarati's UG book....I teach in a UVLRM to people with very limited math background, I guess this course will end up being similar to an advanced undergrad course
He used Gujarati at Grad level? I thought he only had a UG book.
Wooldridge is very good but, as it says on the tin, only has cross sectional and panel metrics.
Wooldridge's undergrad book is very comprehensive. Probably the best UG book around. With regards to the level of math, it is about as tough as you can get without using matrix algebra.